ExternalVendorCode/tahoe-lafs
1
0
Fork 0
mirror of https://github.com/tahoe-lafs/tahoe-lafs.git synced 2025-02-21 02:01:31 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge branch 'master' into 2916.grid-manager-proposal.5

Browse Source
This commit is contained in:
meejah 2020-05-07 18:58:55 -06:00
commit 60fea95e23
327 changed files with 10822 additions and 29987 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
.appveyor.yml
View File

@ -51,6 +51,11 @@ test_script:
# to put the Python version you want to use on PATH.
- |
%PYTHON%\Scripts\tox.exe -e coverage
%PYTHON%\Scripts\tox.exe -e pyinstaller
# To verify that the resultant PyInstaller-generated binary executes
# cleanly (i.e., that it terminates with an exit code of 0 and isn't
# failing due to import/packaging-related errors, etc.).
- dist\Tahoe-LAFS\tahoe.exe --version
after_test:
# This builds the main tahoe wheel, and wheels for all dependencies.
@ -64,7 +69,7 @@ after_test:
%PYTHON%\python.exe setup.py bdist_wheel
%PYTHON%\python.exe -m pip wheel -w dist .
- |
%PYTHON%\python.exe -m pip install codecov coverage
%PYTHON%\python.exe -m pip install codecov "coverage ~= 4.5"
%PYTHON%\python.exe -m coverage xml -o coverage.xml -i
%PYTHON%\python.exe -m codecov -X search -X gcov -f coverage.xml

8
.circleci/Dockerfile.centos
View File

@ -11,11 +11,11 @@ RUN yum install --assumeyes \
git \
sudo \
make automake gcc gcc-c++ \
python \
python-devel \
python2 \
python2-devel \
libffi-devel \
openssl-devel \
libyaml-devel \
libyaml \
/usr/bin/virtualenv \
net-tools
@ -23,4 +23,4 @@ RUN yum install --assumeyes \
# *update* this checkout on each job run, saving us more time per-job.
COPY . ${BUILD_SRC_ROOT}
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}"
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}" "python2.7"

2
.circleci/Dockerfile.debian
View File

@ -23,7 +23,7 @@ RUN apt-get --quiet update && \
# *update* this checkout on each job run, saving us more time per-job.
COPY . ${BUILD_SRC_ROOT}
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}"
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}" "python2.7"
# Only the integration tests currently need this but it doesn't hurt to always
# have it present and it's simpler than building a whole extra image just for

2
.circleci/Dockerfile.fedora
View File

@ -23,4 +23,4 @@ RUN yum install --assumeyes \
# *update* this checkout on each job run, saving us more time per-job.
COPY . ${BUILD_SRC_ROOT}
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}"
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}" "python2.7"

23
.circleci/Dockerfile.pypy Normal file
View File

@ -0,0 +1,23 @@
FROM pypy:2.7-buster
ENV WHEELHOUSE_PATH /tmp/wheelhouse
ENV VIRTUALENV_PATH /tmp/venv
# This will get updated by the CircleCI checkout step.
ENV BUILD_SRC_ROOT /tmp/project
RUN apt-get --quiet update && \
apt-get --quiet --yes install \
git \
lsb-release \
sudo \
build-essential \
libffi-dev \
libssl-dev \
libyaml-dev \
virtualenv
# Get the project source. This is better than it seems. CircleCI will
# *update* this checkout on each job run, saving us more time per-job.
COPY . ${BUILD_SRC_ROOT}
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}" "pypy"

2
.circleci/Dockerfile.slackware
View File

@ -46,4 +46,4 @@ RUN slackpkg install \
# *update* this checkout on each job run, saving us more time per-job.
COPY . ${BUILD_SRC_ROOT}
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}"
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}" "python2.7"

2
.circleci/Dockerfile.ubuntu
View File

@ -26,4 +26,4 @@ RUN apt-get --quiet update && \
# *update* this checkout on each job run, saving us more time per-job.
COPY . ${BUILD_SRC_ROOT}
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}"
RUN "${BUILD_SRC_ROOT}"/.circleci/prepare-image.sh "${WHEELHOUSE_PATH}" "${VIRTUALENV_PATH}" "${BUILD_SRC_ROOT}" "python2.7"

127
.circleci/config.yml
View File

@ -21,12 +21,18 @@ workflows:
requires:
- "fedora-29"
- "centos-7"
- "centos-8"
- "slackware-14.2"
- "nixos-19.09"
# Test against PyPy 2.7
- "pypy2.7-buster"
# Other assorted tasks and configurations
- "lint"
- "pyinstaller"
- "deprecations"
- "c-locale"
# Any locale other than C or UTF-8.
@ -38,6 +44,10 @@ workflows:
# integration tests.
- "debian-9"
# Generate the underlying data for a visualization to aid with Python 3
# porting.
- "build-porting-depgraph"
images:
# Build the Docker images used by the ci jobs. This makes the ci jobs
# faster and takes various spurious failures out of the critical path.
@ -57,8 +67,9 @@ workflows:
- "build-image-ubuntu-18.04"
- "build-image-fedora-28"
- "build-image-fedora-29"
- "build-image-centos-7"
- "build-image-centos-8"
- "build-image-slackware-14.2"
- "build-image-pypy-2.7-buster"
jobs:
@ -79,19 +90,48 @@ jobs:
command: |
~/.local/bin/tox -e codechecks
pyinstaller:
docker:
- image: "circleci/python:2"
steps:
- "checkout"
- run:
name: "Install tox"
command: |
pip install --user tox
- run:
name: "Make PyInstaller executable"
command: |
~/.local/bin/tox -e pyinstaller
- run:
# To verify that the resultant PyInstaller-generated binary executes
# cleanly (i.e., that it terminates with an exit code of 0 and isn't
# failing due to import/packaging-related errors, etc.).
name: "Test PyInstaller executable"
command: |
dist/Tahoe-LAFS/tahoe --version
debian-9: &DEBIAN
docker:
- image: "tahoelafsci/debian:9"
user: "nobody"
environment: &UTF_8_ENVIRONMENT
# In general, the test suite is not allowed to fail while the job
# succeeds. But you can set this to "yes" if you want it to be
# otherwise.
ALLOWED_FAILURE: "no"
# Tell Hypothesis which configuration we want it to use.
TAHOE_LAFS_HYPOTHESIS_PROFILE: "ci"
# Tell the C runtime things about character encoding (mainly to do with
# filenames and argv).
LANG: "en_US.UTF-8"
# Select a tox environment to run for this job.
TAHOE_LAFS_TOX_ENVIRONMENT: "coverage"
TAHOE_LAFS_TOX_ENVIRONMENT: "py27-coverage"
# Additional arguments to pass to tox.
TAHOE_LAFS_TOX_ARGS: ""
# The path in which test artifacts will be placed.
@ -123,6 +163,7 @@ jobs:
/tmp/project/.circleci/run-tests.sh \
"/tmp/venv" \
"/tmp/project" \
"${ALLOWED_FAILURE}" \
"${ARTIFACTS_OUTPUT_PATH}" \
"${TAHOE_LAFS_TOX_ENVIRONMENT}" \
"${TAHOE_LAFS_TOX_ARGS}"
@ -157,6 +198,18 @@ jobs:
user: "nobody"
pypy2.7-buster:
<<: *DEBIAN
docker:
- image: "tahoelafsci/pypy:2.7-buster"
user: "nobody"
environment:
<<: *UTF_8_ENVIRONMENT
TAHOE_LAFS_TOX_ENVIRONMENT: "pypy27-coverage"
ALLOWED_FAILURE: "yes"
c-locale:
<<: *DEBIAN
@ -216,9 +269,9 @@ jobs:
user: "nobody"
centos-7: &RHEL_DERIV
centos-8: &RHEL_DERIV
docker:
- image: "tahoelafsci/centos:7"
- image: "tahoelafsci/centos:8"
user: "nobody"
environment: *UTF_8_ENVIRONMENT
@ -271,6 +324,58 @@ jobs:
- store_artifacts: *STORE_OTHER_ARTIFACTS
- run: *SUBMIT_COVERAGE
nixos-19.09:
docker:
# Run in a highly Nix-capable environment.
- image: "nixorg/nix:circleci"
environment:
NIX_PATH: "nixpkgs=https://github.com/NixOS/nixpkgs-channels/archive/nixos-19.09-small.tar.gz"
steps:
- "checkout"
- "run":
name: "Build and Test"
command: |
# CircleCI build environment looks like it has a zillion and a
# half cores. Don't let Nix autodetect this high core count
# because it blows up memory usage and fails the test run. Pick a
# number of cores that suites the build environment we're paying
# for (the free one!).
#
# Also, let it run more than one job at a time because we have to
# build a couple simple little dependencies that don't take
# advantage of multiple cores and we get a little speedup by doing
# them in parallel.
nix-build --cores 3 --max-jobs 2 nix/
# Generate up-to-date data for the dependency graph visualizer.
build-porting-depgraph:
# Get a system in which we can easily install Tahoe-LAFS and all its
# dependencies. The dependency graph analyzer works by executing the code.
# It's Python, what do you expect?
<<: *DEBIAN
steps:
- "checkout"
- add_ssh_keys:
fingerprints:
# Jean-Paul Calderone <exarkun@twistedmatrix.com> (CircleCI depgraph key)
# This lets us push to tahoe-lafs/tahoe-depgraph in the next step.
- "86:38:18:a7:c0:97:42:43:18:46:55:d6:21:b0:5f:d4"
- run:
name: "Setup Python Environment"
command: |
/tmp/venv/bin/pip install -e /tmp/project
- run:
name: "Generate dependency graph data"
command: |
. /tmp/venv/bin/activate
./misc/python3/depgraph.sh
build-image: &BUILD_IMAGE
# This is a template for a job to build a Docker image that has as much of
# the setup as we can manage already done and baked in. This cuts down on
@ -376,12 +481,12 @@ jobs:
TAG: "18.04"
build-image-centos-7:
build-image-centos-8:
<<: *BUILD_IMAGE
environment:
DISTRO: "centos"
TAG: "7"
TAG: "8"
build-image-fedora-28:
@ -406,3 +511,11 @@ jobs:
environment:
DISTRO: "slackware"
TAG: "14.2"
build-image-pypy-2.7-buster:
<<: *BUILD_IMAGE
environment:
DISTRO: "pypy"
TAG: "2.7-buster"

18
.circleci/create-virtualenv.sh
View File

@ -13,9 +13,14 @@ shift
BOOTSTRAP_VENV="$1"
shift
# The basename of the Python executable (found on PATH) that will be used with
# this image. This lets us create a virtualenv that uses the correct Python.
PYTHON="$1"
shift
# Set up the virtualenv as a non-root user so we can run the test suite as a
# non-root user. See below.
virtualenv --python python2.7 "${BOOTSTRAP_VENV}"
virtualenv --python "${PYTHON}" "${BOOTSTRAP_VENV}"
# For convenience.
PIP="${BOOTSTRAP_VENV}/bin/pip"
@ -33,5 +38,12 @@ export PIP_FIND_LINKS="file://${WHEELHOUSE_PATH}"
"${PIP}" install certifi
# Get a new, awesome version of pip and setuptools. For example, the
# distro-packaged virtualenv's pip may not know about wheels.
"${PIP}" install --upgrade pip setuptools wheel
# distro-packaged virtualenv's pip may not know about wheels. Get the newer
# version of pip *first* in case we have a really old one now which can't even
# install setuptools properly.
"${PIP}" install --upgrade pip
# setuptools 45 requires Python 3.5 or newer. Even though we upgraded pip
# above, it may still not be able to get us a compatible version unless we
# explicitly ask for one.
"${PIP}" install --upgrade setuptools==44.0.0 wheel

2
.circleci/populate-wheelhouse.sh
View File

@ -40,7 +40,7 @@ export PIP_FIND_LINKS="file://${WHEELHOUSE_PATH}"
"${PIP}" \
wheel \
--wheel-dir "${WHEELHOUSE_PATH}" \
"${PROJECT_ROOT}"[test,tor,i2p] \
"${PROJECT_ROOT}"[test] \
${BASIC_DEPS} \
${TEST_DEPS} \
${REPORTING_DEPS}

7
.circleci/prepare-image.sh
View File

@ -18,6 +18,11 @@ shift
PROJECT_ROOT="$1"
shift
# The basename of the Python executable (found on PATH) that will be used with
# this image. This lets us create a virtualenv that uses the correct Python.
PYTHON="$1"
shift
"${PROJECT_ROOT}"/.circleci/fix-permissions.sh "${WHEELHOUSE_PATH}" "${BOOTSTRAP_VENV}" "${PROJECT_ROOT}"
sudo --set-home -u nobody "${PROJECT_ROOT}"/.circleci/create-virtualenv.sh "${WHEELHOUSE_PATH}" "${BOOTSTRAP_VENV}"
sudo --set-home -u nobody "${PROJECT_ROOT}"/.circleci/create-virtualenv.sh "${WHEELHOUSE_PATH}" "${BOOTSTRAP_VENV}" "${PYTHON}"
sudo --set-home -u nobody "${PROJECT_ROOT}"/.circleci/populate-wheelhouse.sh "${WHEELHOUSE_PATH}" "${BOOTSTRAP_VENV}" "${PROJECT_ROOT}"

26
.circleci/run-tests.sh
View File

@ -13,6 +13,9 @@ shift
PROJECT_ROOT="$1"
shift
ALLOWED_FAILURE="$1"
shift
ARTIFACTS=$1
shift
@ -40,6 +43,17 @@ else
JUNITXML=""
fi
# A prefix for the test command that ensure it will exit after no more than a
# certain amount of time. Ideally, we would only enforce a "silent" period
# timeout but there isn't obviously a ready-made tool for that. The test
# suite only takes about 5 - 6 minutes on CircleCI right now. 15 minutes
# seems like a moderately safe window.
#
# This is primarily aimed at catching hangs on the PyPy job which runs for
# about 21 minutes and then gets killed by CircleCI in a way that fails the
# job and bypasses our "allowed failure" logic.
TIMEOUT="timeout --kill-after 1m 15m"
# Run the test suite as a non-root user. This is the expected usage some
# small areas of the test suite assume non-root privileges (such as unreadable
# files being unreadable).
@ -54,14 +68,20 @@ export SUBUNITREPORTER_OUTPUT_PATH="${SUBUNIT2}"
export TAHOE_LAFS_TRIAL_ARGS="--reporter=subunitv2-file --rterrors"
export PIP_NO_INDEX="1"
${BOOTSTRAP_VENV}/bin/tox \
if [ "${ALLOWED_FAILURE}" = "yes" ]; then
alternative="true"
else
alternative="false"
fi
${TIMEOUT} ${BOOTSTRAP_VENV}/bin/tox \
-c ${PROJECT_ROOT}/tox.ini \
--workdir /tmp/tahoe-lafs.tox \
-e "${TAHOE_LAFS_TOX_ENVIRONMENT}" \
${TAHOE_LAFS_TOX_ARGS}
${TAHOE_LAFS_TOX_ARGS} || "${alternative}"
if [ -n "${ARTIFACTS}" ]; then
# Create a junitxml results area.
mkdir -p "$(dirname "${JUNITXML}")"
${BOOTSTRAP_VENV}/bin/subunit2junitxml < "${SUBUNIT2}" > "${JUNITXML}"
${BOOTSTRAP_VENV}/bin/subunit2junitxml < "${SUBUNIT2}" > "${JUNITXML}" || "${alternative}"
fi

2
.coveragerc
View File

@ -8,3 +8,5 @@ source =
omit =
*/allmydata/test/*
*/allmydata/_version.py
parallel = True
branch = True

172
.github/workflows/ci.yml vendored Normal file
View File

@ -0,0 +1,172 @@
name: CI
on:
push:
branches:
- "master"
pull_request:
jobs:
coverage:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
os:
- macos-latest
- windows-latest
python-version:
- 2.7
steps:
# Get vcpython27 on Windows + Python 2.7, to build zfec
# extension. See https://chocolatey.org/packages/vcpython27 and
# https://github.com/crazy-max/ghaction-chocolatey
- name: Install MSVC 9.0 for Python 2.7 [Windows]
if: matrix.os == 'windows-latest' && matrix.python-version == '2.7'
uses: crazy-max/ghaction-chocolatey@v1
with:
args: install vcpython27
- name: Check out Tahoe-LAFS sources
uses: actions/checkout@v2
- name: Fetch all history for all tags and branches
run: git fetch --prune --unshallow
- name: Set up Python ${{ matrix.python-version }}
uses: actions/setup-python@v1
with:
python-version: ${{ matrix.python-version }}
- name: Install Python packages
run: |
pip install --upgrade codecov tox setuptools
pip list
- name: Display tool versions
run: python misc/build_helpers/show-tool-versions.py
- name: Run "tox -e coverage"
run: tox -e coverage
- name: Upload eliot.log in case of failure
uses: actions/upload-artifact@v1
if: failure()
with:
name: eliot.log
path: eliot.log
- name: Upload coverage report
uses: codecov/codecov-action@v1
with:
token: abf679b6-e2e6-4b33-b7b5-6cfbd41ee691
file: coverage.xml
integration:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
os:
- macos-latest
python-version:
- 2.7
steps:
- name: Install Tor [Ubuntu]
if: matrix.os == 'ubuntu-latest'
run: sudo apt install tor
- name: Install Tor [macOS]
if: matrix.os == 'macos-latest'
run: brew install tor
- name: Install Tor [Windows]
if: matrix.os == 'windows-latest'
uses: crazy-max/ghaction-chocolatey@v1
with:
args: install tor
- name: Install MSVC 9.0 for Python 2.7 [Windows]
if: matrix.os == 'windows-latest' && matrix.python-version == '2.7'
uses: crazy-max/ghaction-chocolatey@v1
with:
args: install vcpython27
- name: Check out Tahoe-LAFS sources
uses: actions/checkout@v2
- name: Fetch all history for all tags and branches
run: git fetch --prune --unshallow
- name: Set up Python ${{ matrix.python-version }}
uses: actions/setup-python@v1
with:
python-version: ${{ matrix.python-version }}
- name: Install Python packages
run: |
pip install --upgrade tox
pip list
- name: Display tool versions
run: python misc/build_helpers/show-tool-versions.py
- name: Run "tox -e integration"
run: tox -e integration
- name: Upload eliot.log in case of failure
uses: actions/upload-artifact@v1
if: failure()
with:
name: integration.eliot.json
path: integration.eliot.json
packaging:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
os:
- macos-latest
- windows-latest
- ubuntu-latest
python-version:
- 2.7
steps:
# Get vcpython27 on Windows + Python 2.7, to build zfec
# extension. See https://chocolatey.org/packages/vcpython27 and
# https://github.com/crazy-max/ghaction-chocolatey
- name: Install MSVC 9.0 for Python 2.7 [Windows]
if: matrix.os == 'windows-latest' && matrix.python-version == '2.7'
uses: crazy-max/ghaction-chocolatey@v1
with:
args: install vcpython27
- name: Check out Tahoe-LAFS sources
uses: actions/checkout@v2
- name: Fetch all history for all tags and branches
run: git fetch --prune --unshallow
- name: Set up Python ${{ matrix.python-version }}
uses: actions/setup-python@v1
with:
python-version: ${{ matrix.python-version }}
- name: Install Python packages
run: |
pip install --upgrade codecov tox setuptools
pip list
- name: Display tool versions
run: python misc/build_helpers/show-tool-versions.py
- name: Run "tox -e pyinstaller"
run: tox -e pyinstaller

1
.gitignore vendored
View File

@ -43,7 +43,6 @@ zope.interface-*.egg
/.tox/
/docs/_build/
/coverage.xml
/smoke_magicfolder/
/.hypothesis/
# This is the plaintext of the private environment needed for some CircleCI

6
.travis.yml
View File

@ -25,10 +25,14 @@ script:
- |
set -eo pipefail
if [ "${T}" = "py35" ]; then
python3 -m compileall -f .
python3 -m compileall -f -x tahoe-depgraph.py .
else
tox -e ${T}
fi
# To verify that the resultant PyInstaller-generated binary executes
# cleanly (i.e., that it terminates with an exit code of 0 and isn't
# failing due to import/packaging-related errors, etc.).
if [ "${T}" = "pyinstaller" ]; then dist/Tahoe-LAFS/tahoe --version; fi
after_success:
- if [ "${T}" = "coverage" ]; then codecov; fi

6
Makefile
View File

@ -42,12 +42,6 @@ upload-osx-pkg:
# echo not uploading tahoe-lafs-osx-pkg because this is not trunk but is branch \"${BB_BRANCH}\" ; \
# fi
.PHONY: smoketest
smoketest:
-python ./src/allmydata/test/check_magicfolder_smoke.py kill
-rm -rf smoke_magicfolder/
python ./src/allmydata/test/check_magicfolder_smoke.py
# code coverage-based testing is disabled temporarily, as we switch to tox.
# This will eventually be added to a tox environment. The following comments
# and variable settings are retained as notes for that future effort.

109
NEWS.rst
View File

@ -5,6 +5,115 @@ User-Visible Changes in Tahoe-LAFS
==================================
.. towncrier start line
Release 1.14.0 (2020-03-11)
'''''''''''''''''''''''''''
Features
--------
- Magic-Folders are now supported on macOS. (`#1432 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1432>`_)
- Add a "tox -e draftnews" which runs towncrier in draft mode (`#2942 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2942>`_)
- Fedora 29 is now tested as part of the project's continuous integration system. (`#2955 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2955>`_)
- The Magic-Folder frontend now emits structured, causal logs. This makes it easier for developers to make sense of its behavior and for users to submit useful debugging information alongside problem reports. (`#2972 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2972>`_)
- The `tahoe` CLI now accepts arguments for configuring structured logging messages which Tahoe-LAFS is being converted to emit. This change does not introduce any new defaults for on-filesystem logging. (`#2975 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2975>`_)
- The web API now publishes streaming Eliot logs via a token-protected WebSocket at /private/logs/v1. (`#3006 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3006>`_)
- End-to-end in-memory tests for websocket features (`#3041 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3041>`_)
- allmydata.interfaces.IFoolscapStoragePlugin has been introduced, an extension point for customizing the storage protocol. (`#3049 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3049>`_)
- Static storage server "announcements" in ``private/servers.yaml`` are now individually logged and ignored if they cannot be interpreted. (`#3051 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3051>`_)
- Storage servers can now be configured to load plugins for allmydata.interfaces.IFoolscapStoragePlugin and offer them to clients. (`#3053 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3053>`_)
- Storage clients can now be configured to load plugins for allmydata.interfaces.IFoolscapStoragePlugin and use them to negotiate with servers. (`#3054 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3054>`_)
- The [storage] configuration section now accepts a boolean *anonymous* item to enable or disable anonymous storage access. The default behavior remains unchanged. (`#3184 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3184>`_)
- Enable the helper when creating a node with `tahoe create-node --helper` (`#3235 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3235>`_)
Bug Fixes
---------
- refactor initialization code to be more async-friendly (`#2870 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2870>`_)
- Configuration-checking code wasn't being called due to indenting (`#2935 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2935>`_)
- refactor configuration handling out of Node into _Config (`#2936 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2936>`_)
- "tox -e codechecks" no longer dirties the working tree. (`#2941 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2941>`_)
- Updated the Tor release key, used by the integration tests. (`#2944 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2944>`_)
- `tahoe backup` no longer fails with an unhandled exception when it encounters a special file (device, fifo) in the backup source. (`#2950 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2950>`_)
- Magic-Folders now creates spurious conflict files in fewer cases. In particular, if files are added to the folder while a client is offline, that client will not create conflict files for all those new files when it starts up. (`#2965 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2965>`_)
- The confusing and misplaced sub-command group headings in `tahoe --help` output have been removed. (`#2976 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2976>`_)
- The Magic-Folder frontend is now more responsive to subtree changes on Windows. (`#2997 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2997>`_)
- remove ancient bundled jquery and d3, and the "dowload timeline" feature they support (`#3228 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3228>`_)
Dependency/Installation Changes
-------------------------------
- Tahoe-LAFS no longer makes start-up time assertions about the versions of its dependencies. It is the responsibility of the administrator of the installation to ensure the correct version of dependencies are supplied. (`#2749 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2749>`_)
- Tahoe-LAFS now depends on Twisted 16.6 or newer. (`#2957 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2957>`_)
Removed Features
----------------
- "tahoe rm", an old alias for "tahoe unlink", has been removed. (`#1827 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1827>`_)
- The direct dependencies on pyutil and zbase32 have been removed. (`#2098 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2098>`_)
- Untested and unmaintained code for running Tahoe-LAFS as a Windows service has been removed. (`#2239 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2239>`_)
- The redundant "pypywin32" dependency has been removed. (`#2392 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2392>`_)
- Fedora 27 is no longer tested as part of the project's continuous integration system. (`#2955 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2955>`_)
- "tahoe start", "tahoe daemonize", "tahoe restart", and "tahoe stop" are now deprecated in favor of using "tahoe run", possibly with a third-party process manager. (`#3273 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3273>`_)
Other Changes
-------------
- Tahoe-LAFS now tests for PyPy compatibility on CI. (`#2479 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2479>`_)
- Tahoe-LAFS now requires Twisted 18.4.0 or newer. (`#2771 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2771>`_)
- Tahoe-LAFS now uses towncrier to maintain the NEWS file. (`#2908 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2908>`_)
- The release process document has been updated. (`#2920 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2920>`_)
- allmydata.test.test_system.SystemTest is now more reliable with respect to bound address collisions. (`#2933 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2933>`_)
- The Tox configuration has been fixed to work around a problem on Windows CI. (`#2956 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2956>`_)
- The PyInstaller CI job now works around a pip/pyinstaller incompatibility. (`#2958 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2958>`_)
- Some CI jobs for integration tests have been moved from TravisCI to CircleCI. (`#2959 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2959>`_)
- Several warnings from a new release of pyflakes have been fixed. (`#2960 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2960>`_)
- Some Slackware 14.2 continuous integration problems have been resolved. (`#2961 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2961>`_)
- Some macOS continuous integration failures have been fixed. (`#2962 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2962>`_)
- The NoNetworkGrid implementation has been somewhat improved. (`#2966 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2966>`_)
- A bug in the test suite for the create-alias command has been fixed. (`#2967 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2967>`_)
- The integration test suite has been updated to use pytest-twisted instead of deprecated pytest APIs. (`#2968 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2968>`_)
- The magic-folder integration test suite now performs more aggressive cleanup of the processes it launches. (`#2969 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2969>`_)
- The integration tests now correctly document the `--keep-tempdir` option. (`#2970 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2970>`_)
- A misuse of super() in the integration tests has been fixed. (`#2971 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2971>`_)
- Several utilities to facilitate the use of the Eliot causal logging library have been introduced. (`#2973 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2973>`_)
- The Windows CI configuration has been tweaked. (`#2974 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2974>`_)
- The Magic-Folder frontend has had additional logging improvements. (`#2977 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2977>`_)
- (`#2981 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2981>`_, `#2982 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2982>`_)
- Added a simple sytax checker so that once a file has reached python3 compatibility, it will not regress. (`#3001 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3001>`_)
- Converted all uses of the print statement to the print function in the ./misc/ directory. (`#3002 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3002>`_)
- The contributor guidelines are now linked from the GitHub pull request creation page. (`#3003 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3003>`_)
- Updated the testing code to use the print function instead of the print statement. (`#3008 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3008>`_)
- Replaced print statement with print fuction for all tahoe_* scripts. (`#3009 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3009>`_)
- Replaced all remaining instances of the print statement with the print function. (`#3010 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3010>`_)
- Replace StringIO imports with six.moves. (`#3011 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3011>`_)
- Updated all Python files to use PEP-3110 exception syntax for Python3 compatibility. (`#3013 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3013>`_)
- Update raise syntax for Python3 compatibility. (`#3014 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3014>`_)
- Updated instances of octal literals to use the format 0o123 for Python3 compatibility. (`#3015 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3015>`_)
- allmydata.test.no_network, allmydata.test.test_system, and allmydata.test.web.test_introducer are now more reliable with respect to bound address collisions. (`#3016 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3016>`_)
- Removed tuple unpacking from function and lambda definitions for Python3 compatibility. (`#3019 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3019>`_)
- Updated Python2 long numeric literals for Python3 compatibility. (`#3020 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3020>`_)
- CircleCI jobs are now faster as a result of pre-building configured Docker images for the CI jobs. (`#3024 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3024>`_)
- Removed used of backticks for "repr" for Python3 compatibility. (`#3027 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3027>`_)
- Updated string literal syntax for Python3 compatibility. (`#3028 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3028>`_)
- Updated CI to enforce Python3 syntax for entire repo. (`#3030 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3030>`_)
- Replaced pycryptopp with cryptography. (`#3031 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3031>`_)
- All old-style classes ported to new-style. (`#3042 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3042>`_)
- Whitelisted "/bin/mv" as command for codechecks performed by tox. This fixes a current warning and prevents future errors (for tox 4). (`#3043 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3043>`_)
- Progress towards Python 3 compatibility is now visible at <https://tahoe-lafs.github.io/tahoe-depgraph/>. (`#3152 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3152>`_)
- Collect coverage information from integration tests (`#3234 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3234>`_)
- NixOS is now a supported Tahoe-LAFS platform. (`#3266 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3266>`_)
Misc/Other
----------
- `#1893 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1893>`_, `#2266 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2266>`_, `#2283 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2283>`_, `#2766 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2766>`_, `#2980 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2980>`_, `#2985 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2985>`_, `#2986 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2986>`_, `#2987 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2987>`_, `#2988 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2988>`_, `#2989 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2989>`_, `#2990 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2990>`_, `#2991 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2991>`_, `#2992 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2992>`_, `#2995 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2995>`_, `#3000 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3000>`_, `#3004 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3004>`_, `#3005 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3005>`_, `#3007 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3007>`_, `#3012 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3012>`_, `#3017 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3017>`_, `#3021 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3021>`_, `#3023 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3023>`_, `#3025 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3025>`_, `#3026 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3026>`_, `#3029 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3029>`_, `#3036 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3036>`_, `#3038 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3038>`_, `#3048 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3048>`_, `#3086 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3086>`_, `#3097 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3097>`_, `#3111 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3111>`_, `#3118 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3118>`_, `#3119 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3119>`_, `#3227 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3227>`_, `#3229 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3229>`_, `#3232 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3232>`_, `#3233 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3233>`_, `#3237 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3237>`_, `#3238 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3238>`_, `#3239 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3239>`_, `#3240 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3240>`_, `#3242 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3242>`_, `#3243 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3243>`_, `#3245 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3245>`_, `#3246 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3246>`_, `#3248 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3248>`_, `#3250 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3250>`_, `#3252 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3252>`_, `#3255 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3255>`_, `#3256 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3256>`_, `#3259 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3259>`_, `#3261 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3261>`_, `#3262 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3262>`_, `#3263 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3263>`_, `#3264 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3264>`_, `#3265 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3265>`_, `#3267 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3267>`_, `#3268 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3268>`_, `#3271 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3271>`_, `#3272 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3272>`_, `#3274 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3274>`_, `#3275 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3275>`_, `#3276 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3276>`_, `#3279 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3279>`_, `#3281 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3281>`_, `#3282 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3282>`_, `#3285 <https://tahoe-lafs.org/trac/tahoe-lafs/ticket/3285>`_
Release 1.13.0 (05-August-2018)
'''''''''''''''''''''''''''''''

30
docs/INSTALL.rst
View File

@ -68,6 +68,8 @@ compile the dependencies yourself (instead of using ``--find-links`` to take
advantage of the pre-compiled ones we host), you'll also need to install
Xcode and its command-line tools.
**Note** that Tahoe-LAFS depends on `openssl 1.1.1c` or greater.
Python 2.7
----------
@ -121,6 +123,9 @@ On Debian/Ubuntu-derived systems, the necessary packages are ``python-dev``,
RPM-based system (like Fedora) these may be named ``python-devel``, etc,
instead, and cam be installed with ``yum`` or ``rpm``.
**Note** that Tahoe-LAFS depends on `openssl 1.1.1c` or greater.
Install the Latest Tahoe-LAFS Release
=====================================
@ -158,7 +163,7 @@ from PyPI with ``venv/bin/pip install tahoe-lafs``. After installation, run
Successfully installed ...
% venv/bin/tahoe --version
tahoe-lafs: 1.13.0
tahoe-lafs: 1.14.0
foolscap: ...
%
@ -178,16 +183,27 @@ You can also install directly from the source tarball URL::
New python executable in ~/venv/bin/python2.7
Installing setuptools, pip, wheel...done.
% venv/bin/pip install https://tahoe-lafs.org/downloads/tahoe-lafs-1.13.0.tar.bz2
Collecting https://tahoe-lafs.org/downloads/tahoe-lafs-1.13.0.tar.bz2
% venv/bin/pip install https://tahoe-lafs.org/downloads/tahoe-lafs-1.14.0.tar.bz2
Collecting https://tahoe-lafs.org/downloads/tahoe-lafs-1.14.0.tar.bz2
...
Installing collected packages: ...
Successfully installed ...
% venv/bin/tahoe --version
tahoe-lafs: 1.13.0
tahoe-lafs: 1.14.0
...
Extras
------
Tahoe-LAFS provides some functionality only when explicitly requested at installation time.
It does this using the "extras" feature of setuptools.
You can request these extra features when running the ``pip install`` command like this::
% venv/bin/pip install tahoe-lafs[tor]
This example enables support for listening and connecting using Tor.
The Tahoe-LAFS documentation for specific features which require an explicit install-time step will mention the "extra" that must be requested.
Hacking On Tahoe-LAFS
---------------------
@ -208,7 +224,7 @@ the additional libraries needed to run the unit tests::
Successfully installed ...
% venv/bin/tahoe --version
tahoe-lafs: 1.13.0.post34.dev0
tahoe-lafs: 1.14.0.post34.dev0
...
This way, you won't have to re-run the ``pip install`` step each time you
@ -257,7 +273,7 @@ result in a "all tests passed" mesage::
% tox
GLOB sdist-make: ~/tahoe-lafs/setup.py
py27 recreate: ~/tahoe-lafs/.tox/py27
py27 inst: ~/tahoe-lafs/.tox/dist/tahoe-lafs-1.13.0.post8.dev0.zip
py27 inst: ~/tahoe-lafs/.tox/dist/tahoe-lafs-1.14.0.post8.dev0.zip
py27 runtests: commands[0] | tahoe --version
py27 runtests: commands[1] | trial --rterrors allmydata
allmydata.test.test_auth
@ -284,6 +300,8 @@ Similar errors about ``openssl/crypto.h`` indicate that you are missing the
OpenSSL development headers (``libssl-dev``). Likewise ``ffi.h`` means you
need ``libffi-dev``.
**Note** that Tahoe-LAFS depends on `openssl 1.1.1c` or greater.
Using Tahoe-LAFS
================

2
docs/architecture.rst
View File

@ -98,7 +98,7 @@ subset are needed to reconstruct the segment (3 out of 10, with the default
settings).
It sends one block from each segment to a given server. The set of blocks on
a given server constitutes a "share". Therefore a subset f the shares (3 out
a given server constitutes a "share". Therefore a subset of the shares (3 out
of 10, by default) are needed to reconstruct the file.
A hash of the encryption key is used to form the "storage index", which is

46
docs/configuration.rst
View File

@ -9,6 +9,7 @@ Configuring a Tahoe-LAFS node
#. `Connection Management`_
#. `Client Configuration`_
#. `Storage Server Configuration`_
#. `Storage Server Plugin Configuration`_
#. `Frontend Configuration`_
#. `Running A Helper`_
#. `Running An Introducer`_
@ -81,7 +82,6 @@ Client/server nodes provide one or more of the following services:
* web-API service
* SFTP service
* FTP service
* Magic Folder service
* helper service
* storage service.
@ -718,12 +718,6 @@ SFTP, FTP
for instructions on configuring these services, and the ``[sftpd]`` and
``[ftpd]`` sections of ``tahoe.cfg``.
Magic Folder
A node running on Linux or Windows can be configured to automatically
upload files that are created or changed in a specified local directory.
See :doc:`frontends/magic-folder` for details.
Storage Server Configuration
============================
@ -738,6 +732,17 @@ Storage Server Configuration
for clients who do not wish to provide storage service. The default value
is ``True``.
``anonymous = (boolean, optional)``
If this is ``True``, the node will expose the storage server via Foolscap
without any additional authentication or authorization. The capability to
use all storage services is conferred by knowledge of the Foolscap fURL
for the storage server which will be included in the storage server's
announcement. If it is ``False``, the node will not expose this and
storage must be exposed using the storage server plugin system (see
`Storage Server Plugin Configuration`_ for details). The default value is
``True``.
``readonly = (boolean, optional)``
If ``True``, the node will run a storage server but will not accept any
@ -798,6 +803,33 @@ Storage Server Configuration
In addition,
see :doc:`accepting-donations` for a convention encouraging donations to storage server operators.
Storage Server Plugin Configuration
===================================
In addition to the built-in storage server,
it is also possible to load and configure storage server plugins into Tahoe-LAFS.
Plugins to load are specified in the ``[storage]`` section.
``plugins = (string, optional)``
This gives a comma-separated list of plugin names.
Plugins named here will be loaded and offered to clients.
The default is for no such plugins to be loaded.
Each plugin can also be configured in a dedicated section.
The section for each plugin is named after the plugin itself::
[storageserver.plugins.<plugin name>]
For example,
the configuration section for a plugin named ``acme-foo-v1`` is ``[storageserver.plugins.acme-foo-v1]``.
The contents of such sections are defined by the plugins themselves.
Refer to the documentation provided with those plugins.
Running A Helper
================

4
docs/debian.rst
View File

@ -39,16 +39,16 @@ virtualenv.
The ``.deb`` packages, of course, rely solely upon other ``.deb`` packages.
For reference, here is a list of the debian package names that provide Tahoe's
dependencies as of the 1.9 release:
dependencies as of the 1.14.0 release:
* python
* python-zfec
* python-pycryptopp
* python-foolscap
* python-openssl (needed by foolscap)
* python-twisted
* python-nevow
* python-mock
* python-cryptography
* python-simplejson
* python-setuptools
* python-support (for Debian-specific install-time tools)

2
docs/frontends/CLI.rst
View File

@ -44,7 +44,7 @@ arguments. "``tahoe --help``" might also provide something useful.
Running "``tahoe --version``" will display a list of version strings, starting
with the "allmydata" module (which contains the majority of the Tahoe-LAFS
functionality) and including versions for a number of dependent libraries,
like Twisted, Foolscap, pycryptopp, and zfec. "``tahoe --version-and-path``"
like Twisted, Foolscap, cryptography, and zfec. "``tahoe --version-and-path``"
will also show the path from which each library was imported.
On Unix systems, the shell expands filename wildcards (``'*'`` and ``'?'``)

9
docs/frontends/FTP-and-SFTP.rst
View File

@ -211,14 +211,7 @@ Dependencies
The Tahoe-LAFS SFTP server requires the Twisted "Conch" component (a "conch"
is a twisted shell, get it?). Many Linux distributions package the Conch code
separately: debian puts it in the "python-twisted-conch" package. Conch
requires the "pycrypto" package, which is a Python+C implementation of many
cryptographic functions (the debian package is named "python-crypto").
Note that "pycrypto" is different than the "pycryptopp" package that
Tahoe-LAFS uses (which is a Python wrapper around the C++ -based Crypto++
library, a library that is frequently installed as /usr/lib/libcryptopp.a, to
avoid problems with non-alphanumerics in filenames).
separately: debian puts it in the "python-twisted-conch" package.
Immutable and Mutable Files
===========================

148
docs/frontends/magic-folder.rst
View File

@ -1,148 +0,0 @@
.. -*- coding: utf-8-with-signature -*-
================================
Tahoe-LAFS Magic Folder Frontend
================================
1. `Introduction`_
2. `Configuration`_
3. `Known Issues and Limitations With Magic-Folder`_
Introduction
============
The Magic Folder frontend synchronizes local directories on two or more
clients, using a Tahoe-LAFS grid for storage. Whenever a file is created
or changed under the local directory of one of the clients, the change is
propagated to the grid and then to the other clients.
The implementation of the "drop-upload" frontend, on which Magic Folder is
based, was written as a prototype at the First International Tahoe-LAFS
Summit in June 2011. In 2015, with the support of a grant from the
`Open Technology Fund`_, it was redesigned and extended to support
synchronization between clients. It currently works on Linux and Windows.
Magic Folder is not currently in as mature a state as the other frontends
(web, CLI, SFTP and FTP). This means that you probably should not rely on
all changes to files in the local directory to result in successful uploads.
There might be (and have been) incompatible changes to how the feature is
configured.
We are very interested in feedback on how well this feature works for you, and
suggestions to improve its usability, functionality, and reliability.
.. _`Open Technology Fund`: https://www.opentech.fund/
Configuration
=============
The Magic Folder frontend runs as part of a gateway node. To set it up, you
must use the tahoe magic-folder CLI. For detailed information see our
:doc:`Magic-Folder CLI design
documentation<../proposed/magic-folder/user-interface-design>`. For a
given Magic-Folder collective directory you need to run the ``tahoe
magic-folder create`` command. After that the ``tahoe magic-folder invite``
command must used to generate an *invite code* for each member of the
magic-folder collective. A confidential, authenticated communications channel
should be used to transmit the invite code to each member, who will be
joining using the ``tahoe magic-folder join`` command.
These settings are persisted in the ``[magic_folder]`` section of the
gateway's ``tahoe.cfg`` file.
``[magic_folder]``
``enabled = (boolean, optional)``
If this is ``True``, Magic Folder will be enabled. The default value is
``False``.
``local.directory = (UTF-8 path)``
This specifies the local directory to be monitored for new or changed
files. If the path contains non-ASCII characters, it should be encoded
in UTF-8 regardless of the system's filesystem encoding. Relative paths
will be interpreted starting from the node's base directory.
You should not normally need to set these fields manually because they are
set by the ``tahoe magic-folder create`` and/or ``tahoe magic-folder join``
commands. Use the ``--help`` option to these commands for more information.
After setting up a Magic Folder collective and starting or restarting each
gateway, you can confirm that the feature is working by copying a file into
any local directory, and checking that it appears on other clients.
Large files may take some time to appear.
The 'Operational Statistics' page linked from the Welcome page shows counts
of the number of files uploaded, the number of change events currently
queued, and the number of failed uploads. The 'Recent Uploads and Downloads'
page and the node :doc:`log<../logging>` may be helpful to determine the
cause of any failures.
.. _Known Issues in Magic-Folder:
Known Issues and Limitations With Magic-Folder
==============================================
This feature only works on Linux and Windows. There is a ticket to add
support for Mac OS X and BSD-based systems (`#1432`_).
The only way to determine whether uploads have failed is to look at the
'Operational Statistics' page linked from the Welcome page. This only shows
a count of failures, not the names of files. Uploads are never retried.
The Magic Folder frontend performs its uploads sequentially (i.e. it waits
until each upload is finished before starting the next), even when there
would be enough memory and bandwidth to efficiently perform them in parallel.
A Magic Folder upload can occur in parallel with an upload by a different
frontend, though. (`#1459`_)
On Linux, if there are a large number of near-simultaneous file creation or
change events (greater than the number specified in the file
``/proc/sys/fs/inotify/max_queued_events``), it is possible that some events
could be missed. This is fairly unlikely under normal circumstances, because
the default value of ``max_queued_events`` in most Linux distributions is
16384, and events are removed from this queue immediately without waiting for
the corresponding upload to complete. (`#1430`_)
The Windows implementation might also occasionally miss file creation or
change events, due to limitations of the underlying Windows API
(ReadDirectoryChangesW). We do not know how likely or unlikely this is.
(`#1431`_)
Some filesystems may not support the necessary change notifications.
So, it is recommended for the local directory to be on a directly attached
disk-based filesystem, not a network filesystem or one provided by a virtual
machine.
The ``private/magic_folder_dircap`` and ``private/collective_dircap`` files
cannot use an alias or path to specify the upload directory. (`#1711`_)
If a file in the upload directory is changed (actually relinked to a new
file), then the old file is still present on the grid, and any other caps
to it will remain valid. Eventually it will be possible to use
:doc:`../garbage-collection` to reclaim the space used by these files; however
currently they are retained indefinitely. (`#2440`_)
Unicode filenames are supported on both Linux and Windows, but on Linux, the
local name of a file must be encoded correctly in order for it to be uploaded.
The expected encoding is that printed by
``python -c "import sys; print sys.getfilesystemencoding()"``.
On Windows, local directories with non-ASCII names are not currently working.
(`#2219`_)
On Windows, when a node has Magic Folder enabled, it is unresponsive to Ctrl-C
(it can only be killed using Task Manager or similar). (`#2218`_)
.. _`#1430`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1430
.. _`#1431`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1431
.. _`#1432`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1432
.. _`#1459`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1459
.. _`#1711`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1711
.. _`#2218`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2218
.. _`#2219`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2219
.. _`#2440`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2440

19
docs/historical/historical_known_issues.txt
View File

@ -272,22 +272,3 @@ that size, assume that they have been corrupted and are not retrievable from the
Tahoe storage grid. Tahoe v1.1 clients will refuse to upload files larger than
12 GiB with a clean failure. A future release of Tahoe will remove this
limitation so that larger files can be uploaded.
=== pycryptopp defect resulting in data corruption ===
Versions of pycryptopp earlier than pycryptopp-0.5.0 had a defect
which, when compiled with some compilers, would cause AES-256
encryption and decryption to be computed incorrectly. This could
cause data corruption. Tahoe v1.0 required, and came with a bundled
copy of, pycryptopp v0.3.
==== how to manage it ====
You can detect whether pycryptopp-0.3 has this failure when it is
compiled by your compiler. Run the unit tests that come with
pycryptopp-0.3: unpack the "pycryptopp-0.3.tar" file that comes in the
Tahoe v1.0 {{{misc/dependencies}}} directory, cd into the resulting
{{{pycryptopp-0.3.0}}} directory, and execute {{{python ./setup.py
test}}}. If the tests pass, then your compiler does not trigger this
failure.

7
docs/how_to_make_a_tahoe-lafs_release.org
View File

@ -17,13 +17,14 @@ people are Release Maintainers:
- [ ] all appveyor checks pass
- [ ] all buildbot workers pass their checks
* freeze master branch [0/]
* freeze master branch [0/1]
- [ ] announced the freeze of the master branch on IRC (i.e. non-release PRs won't be merged until after release)
* sync documentation [0/7]
- [ ] NEWS.rst: summarize user-visible changes, aim for one page of text
- [ ] NEWS.rst: (run "tox -e news")
- [ ] added final release name and date to top-most item in NEWS.rst
- [ ] updated relnotes.txt
- [ ] updated relnotes.txt (change next, last versions; summarize NEWS)
- [ ] updated CREDITS
- [ ] updated docs/known_issues.rst
- [ ] docs/INSTALL.rst only points to current tahoe-lafs-X.Y.Z.tar.gz source code file

2
docs/index.rst
View File

@ -20,7 +20,6 @@ Contents:
frontends/CLI
frontends/webapi
frontends/FTP-and-SFTP
frontends/magic-folder
frontends/download-status
known_issues
@ -37,7 +36,6 @@ Contents:
expenses
cautions
write_coordination
magic-folder-howto
backupdb
anonymity-configuration

176
docs/magic-folder-howto.rst
View File

@ -1,176 +0,0 @@
.. _magic-folder-howto:
=========================
Magic Folder Set-up Howto
=========================
#. `This document`_
#. `Setting up a local test grid`_
#. `Setting up Magic Folder`_
#. `Testing`_
This document
=============
This is preliminary documentation of how to set up Magic Folder using a test
grid on a single Linux or Windows machine, with two clients and one server.
It is aimed at a fairly technical audience.
For an introduction to Magic Folder and how to configure it
more generally, see :doc:`frontends/magic-folder`.
It it possible to adapt these instructions to run the nodes on
different machines, to synchronize between three or more clients,
to mix Windows and Linux clients, and to use multiple servers
(if the Tahoe-LAFS encoding parameters are changed).
Setting up a local test grid
============================
Linux
-----
Run these commands::
mkdir ../grid
bin/tahoe create-introducer ../grid/introducer
bin/tahoe start ../grid/introducer
export FURL=`cat ../grid/introducer/private/introducer.furl`
bin/tahoe create-node --introducer="$FURL" ../grid/server
bin/tahoe create-client --introducer="$FURL" ../grid/alice
bin/tahoe create-client --introducer="$FURL" ../grid/bob
Windows
-------
Run::
mkdir ..\grid
bin\tahoe create-introducer ..\grid\introducer
bin\tahoe start ..\grid\introducer
Leave the introducer running in that Command Prompt,
and in a separate Command Prompt (with the same current
directory), run::
set /p FURL=<..\grid\introducer\private\introducer.furl
bin\tahoe create-node --introducer=%FURL% ..\grid\server
bin\tahoe create-client --introducer=%FURL% ..\grid\alice
bin\tahoe create-client --introducer=%FURL% ..\grid\bob
Both Linux and Windows
----------------------
(Replace ``/`` with ``\`` for Windows paths.)
Edit ``../grid/alice/tahoe.cfg``, and make the following
changes to the ``[node]`` and ``[client]`` sections::
[node]
nickname = alice
web.port = tcp:3457:interface=127.0.0.1
[client]
shares.needed = 1
shares.happy = 1
shares.total = 1
Edit ``../grid/bob/tahoe.cfg``, and make the following
change to the ``[node]`` section, and the same change as
above to the ``[client]`` section::
[node]
nickname = bob
web.port = tcp:3458:interface=127.0.0.1
Note that when running nodes on a single machine,
unique port numbers must be used for each node (and they
must not clash with ports used by other server software).
Here we have used the default of 3456 for the server,
3457 for alice, and 3458 for bob.
Now start all of the nodes (the introducer should still be
running from above)::
bin/tahoe start ../grid/server
bin/tahoe start ../grid/alice
bin/tahoe start ../grid/bob
On Windows, a separate Command Prompt is needed to run each
node.
Open a web browser on http://127.0.0.1:3457/ and verify that
alice is connected to the introducer and one storage server.
Then do the same for http://127.0.0.1:3568/ to verify that
bob is connected. Leave all of the nodes running for the
next stage.
Setting up Magic Folder
=======================
Linux
-----
Run::
mkdir -p ../local/alice ../local/bob
bin/tahoe -d ../grid/alice magic-folder create magic: alice ../local/alice
bin/tahoe -d ../grid/alice magic-folder invite magic: bob >invitecode
export INVITECODE=`cat invitecode`
bin/tahoe -d ../grid/bob magic-folder join "$INVITECODE" ../local/bob
bin/tahoe restart ../grid/alice
bin/tahoe restart ../grid/bob
Windows
-------
Run::
mkdir ..\local\alice ..\local\bob
bin\tahoe -d ..\grid\alice magic-folder create magic: alice ..\local\alice
bin\tahoe -d ..\grid\alice magic-folder invite magic: bob >invitecode
set /p INVITECODE=<invitecode
bin\tahoe -d ..\grid\bob magic-folder join %INVITECODE% ..\local\bob
Then close the Command Prompt windows that are running the alice and bob
nodes, and open two new ones in which to run::
bin\tahoe start ..\grid\alice
bin\tahoe start ..\grid\bob
Testing
=======
You can now experiment with creating files and directories in
``../local/alice`` and ``/local/bob``; any changes should be
propagated to the other directory.
Note that when a file is deleted, the corresponding file in the
other directory will be renamed to a filename ending in ``.backup``.
Deleting a directory will have no effect.
For other known issues and limitations, see :ref:`Known Issues in
Magic-Folder`.
As mentioned earlier, it is also possible to run the nodes on
different machines, to synchronize between three or more clients,
to mix Windows and Linux clients, and to use multiple servers
(if the Tahoe-LAFS encoding parameters are changed).
Configuration
=============
There will be a ``[magic_folder]`` section in your ``tahoe.cfg`` file
after setting up Magic Folder.
There is an option you can add to this called ``poll_interval=`` to
control how often (in seconds) the Downloader will check for new things
to download.

4
docs/magic-wormhole-invites.rst
View File

@ -19,9 +19,7 @@ Invites and Joins
Inside Tahoe-LAFS we are using a channel created using `magic
wormhole`_ to exchange configuration and the secret fURL of the
Introducer with new clients. In the future, we would like to make the
Magic Folder (:ref:`Magic Folder HOWTO <magic-folder-howto>`) invites and joins work this way
as well.
Introducer with new clients.
This is a two-part process. Alice runs a grid and wishes to have her
friend Bob use it as a client. She runs ``tahoe invite bob`` which

19
docs/proposed/accounting-overview.txt
View File

@ -546,16 +546,15 @@ The "restrictions dictionary" is a table which establishes an upper bound on
how this authority (or any attenuations thereof) may be used. It is
effectively a set of key-value pairs.
A "signing key" is an EC-DSA192 private key string, as supplied to the
pycryptopp SigningKey() constructor, and is 12 bytes long. A "verifying key"
is an EC-DSA192 public key string, as produced by pycryptopp, and is 24 bytes
long. A "key identifier" is a string which securely identifies a specific
signing/verifying keypair: for long RSA keys it would be a secure hash of the
public key, but since ECDSA192 keys are so short, we simply use the full
verifying key verbatim. A "key hint" is a variable-length prefix of the key
identifier, perhaps zero bytes long, used to help a recipient reduce the
number of verifying keys that it must search to find one that matches a
signed message.
A "signing key" is an EC-DSA192 private key string and is 12 bytes
long. A "verifying key" is an EC-DSA192 public key string, and is 24
bytes long. A "key identifier" is a string which securely identifies a
specific signing/verifying keypair: for long RSA keys it would be a
secure hash of the public key, but since ECDSA192 keys are so short,
we simply use the full verifying key verbatim. A "key hint" is a
variable-length prefix of the key identifier, perhaps zero bytes long,
used to help a recipient reduce the number of verifying keys that it
must search to find one that matches a signed message.
==== Authority Chains ====

4
docs/proposed/index.rst
View File

@ -14,8 +14,4 @@ index only lists the files that are in .rst format.
:maxdepth: 2
leasedb
magic-folder/filesystem-integration
magic-folder/remote-to-local-sync
magic-folder/user-interface-design
magic-folder/multi-party-conflict-detection
http-storage-node-protocol

118
docs/proposed/magic-folder/filesystem-integration.rst
View File

@ -1,118 +0,0 @@
Magic Folder local filesystem integration design
================================================
*Scope*
This document describes how to integrate the local filesystem with Magic
Folder in an efficient and reliable manner. For now we ignore Remote to
Local synchronization; the design and implementation of this is scheduled
for a later time. We also ignore multiple writers for the same Magic
Folder, which may or may not be supported in future. The design here will
be updated to account for those features in later Objectives. Objective 3
may require modifying the database schema or operation, and Objective 5
may modify the User interface.
Tickets on the Tahoe-LAFS trac with the `otf-magic-folder-objective2`_
keyword are within the scope of the local filesystem integration for
Objective 2.
.. _otf-magic-folder-objective2: https://tahoe-lafs.org/trac/tahoe-lafs/query?status=!closed&keywords=~otf-magic-folder-objective2
.. _filesystem_integration-local-scanning-and-database:
*Local scanning and database*
When a Magic-Folder-enabled node starts up, it scans all directories
under the local directory and adds every file to a first-in first-out
"scan queue". When processing the scan queue, redundant uploads are
avoided by using the same mechanism the Tahoe backup command uses: we
keep track of previous uploads by recording each file's metadata such as
size, ``ctime`` and ``mtime``. This information is stored in a database,
referred to from now on as the magic folder db. Using this recorded
state, we ensure that when Magic Folder is subsequently started, the
local directory tree can be scanned quickly by comparing current
filesystem metadata with the previously recorded metadata. Each file
referenced in the scan queue is uploaded only if its metadata differs at
the time it is processed. If a change event is detected for a file that
is already queued (and therefore will be processed later), the redundant
event is ignored.
To implement the magic folder db, we will use an SQLite schema that
initially is the existing Tahoe-LAFS backup schema. This schema may
change in later objectives; this will cause no backward compatibility
problems, because this new feature will be developed on a branch that
makes no compatibility guarantees. However we will have a separate SQLite
database file and separate mutex lock just for Magic Folder. This avoids
usability problems related to mutual exclusion. (If a single file and
lock were used, a backup would block Magic Folder updates for a long
time, and a user would not be able to tell when backups are possible
because Magic Folder would acquire a lock at arbitrary times.)
*Eventual consistency property*
During the process of reading a file in order to upload it, it is not
possible to prevent further local writes. Such writes will result in
temporary inconsistency (that is, the uploaded file will not reflect
what the contents of the local file were at any specific time). Eventual
consistency is reached when the queue of pending uploads is empty. That
is, a consistent snapshot will be achieved eventually when local writes
to the target folder cease for a sufficiently long period of time.
*Detecting filesystem changes*
For the Linux implementation, we will use the `inotify`_ Linux kernel
subsystem to gather events on the local Magic Folder directory tree. This
implementation was already present in Tahoe-LAFS 1.9.0, but needs to be
changed to gather directory creation and move events, in addition to the
events indicating that a file has been written that are gathered by the
current code.
.. _`inotify`: https://en.wikipedia.org/wiki/Inotify
For the Windows implementation, we will use the ``ReadDirectoryChangesW``
Win32 API. The prototype implementation simulates a Python interface to
the inotify API in terms of ``ReadDirectoryChangesW``, allowing most of
the code to be shared across platforms.
The alternative of using `NTFS Change Journals`_ for Windows was
considered, but appears to be more complicated and does not provide any
additional functionality over the scanning approach described above.
The Change Journal mechanism is also only available for NTFS filesystems,
but FAT32 filesystems are still common in user installations of Windows.
.. _`NTFS Change Journals`: https://msdn.microsoft.com/en-us/library/aa363803%28VS.85%29.aspx
When we detect the creation of a new directory below the local Magic
Folder directory, we create it in the Tahoe-LAFS filesystem, and also
scan the new local directory for new files. This scan is necessary to
avoid missing events for creation of files in a new directory before it
can be watched, and to correctly handle cases where an existing directory
is moved to be under the local Magic Folder directory.
*User interface*
The Magic Folder local filesystem integration will initially have a
provisional configuration file-based interface that may not be ideal from
a usability perspective. Creating our local filesystem integration in
this manner will allow us to use and test it independently of the rest of
the Magic Folder software components. We will focus greater attention on
user interface design as a later milestone in our development roadmap.
The configuration file, ``tahoe.cfg``, must define a target local
directory to be synchronized. Provisionally, this configuration will
replace the current ``[drop_upload]`` section::
[magic_folder]
enabled = true
local.directory = "/home/human"
When a filesystem directory is first configured for Magic Folder, the user
needs to create the remote Tahoe-LAFS directory using ``tahoe mkdir``,
and configure the Magic-Folder-enabled node with its URI (e.g. by putting
it in a file ``private/magic_folder_dircap``). If there are existing
files in the local directory, they will be uploaded as a result of the
initial scan described earlier.

373
docs/proposed/magic-folder/multi-party-conflict-detection.rst
View File

@ -1,373 +0,0 @@
Multi-party Conflict Detection
==============================
The current Magic-Folder remote conflict detection design does not properly detect remote conflicts
for groups of three or more parties. This design is specified in the "Fire Dragon" section of this document:
https://github.com/tahoe-lafs/tahoe-lafs/blob/2551.wip.2/docs/proposed/magic-folder/remote-to-local-sync.rst#fire-dragons-distinguishing-conflicts-from-overwrites
This Tahoe-LAFS trac ticket comment outlines a scenario with
three parties in which a remote conflict is falsely detected:
.. _`ticket comment`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2551#comment:22
Summary and definitions
=======================
Abstract file: a file being shared by a Magic Folder.
Local file: a file in a client's local filesystem corresponding to an abstract file.
Relative path: the path of an abstract or local file relative to the Magic Folder root.
Version: a snapshot of an abstract file, with associated metadata, that is uploaded by a Magic Folder client.
A version is associated with the file's relative path, its contents, and
mtime and ctime timestamps. Versions also have a unique identity.
Follows relation:
* If and only if a change to a client's local file at relative path F that results in an upload of version V',
was made when the client already had version V of that file, then we say that V' directly follows V.
* The follows relation is the irreflexive transitive closure of the "directly follows" relation.
The follows relation is transitive and acyclic, and therefore defines a DAG called the
Version DAG. Different abstract files correspond to disconnected sets of nodes in the Version DAG
(in other words there are no "follows" relations between different files).
The DAG is only ever extended, not mutated.
The desired behaviour for initially classifying overwrites and conflicts is as follows:
* if a client Bob currently has version V of a file at relative path F, and it sees a new version V'
of that file in another client Alice's DMD, such that V' follows V, then the write of the new version
is initially an overwrite and should be to the same filename.
* if, in the same situation, V' does not follow V, then the write of the new version should be
classified as a conflict.
The existing :doc:`remote-to-local-sync` document defines when an initial
overwrite should be reclassified as a conflict.
The above definitions completely specify the desired solution of the false
conflict behaviour described in the `ticket comment`_. However, they do not give
a concrete algorithm to compute the follows relation, or a representation in the
Tahoe-LAFS file store of the metadata needed to compute it.
We will consider two alternative designs, proposed by Leif Ryge and
Zooko Wilcox-O'Hearn, that aim to fill this gap.
Leif's Proposal: Magic-Folder "single-file" snapshot design
===========================================================
Abstract
--------
We propose a relatively simple modification to the initial Magic Folder design which
adds merkle DAGs of immutable historical snapshots for each file. The full history
does not necessarily need to be retained, and the choice of how much history to retain
can potentially be made on a per-file basis.
Motivation:
-----------
no SPOFs, no admins
```````````````````
Additionally, the initial design had two cases of excess authority:
1. The magic folder administrator (inviter) has everyone's write-caps and is thus essentially "root"
2. Each client shares ambient authority and can delete anything or everything and
(assuming there is not a conflict) the data will be deleted from all clients. So, each client
is effectively "root" too.
Thus, while it is useful for file synchronization, the initial design is a much less safe place
to store data than in a single mutable tahoe directory (because more client computers have the
possibility to delete it).
Glossary
--------
- merkle DAG: like a merkle tree but with multiple roots, and with each node potentially having multiple parents
- magic folder: a logical directory that can be synchronized between many clients
(devices, users, ...) using a Tahoe-LAFS storage grid
- client: a Magic-Folder-enabled Tahoe-LAFS client instance that has access to a magic folder
- DMD: "distributed mutable directory", a physical Tahoe-LAFS mutable directory.
Each client has the write cap to their own DMD, and read caps to all other client's DMDs
(as in the original Magic Folder design).
- snapshot: a reference to a version of a file; represented as an immutable directory containing
an entry called "content" (pointing to the immutable file containing the file's contents),
and an entry called "parent0" (pointing to a parent snapshot), and optionally parent1 through
parentN pointing at other parents. The Magic Folder snapshot object is conceptually very similar
to a git commit object, except for that it is created automatically and it records the history of an
individual file rather than an entire repository. Also, commits do not need to have authors
(although an author field could be easily added later).
- deletion snapshot: immutable directory containing no content entry (only one or more parents)
- capability: a Tahoe-LAFS diminishable cryptographic capability
- cap: short for capability
- conflict: the situation when another client's current snapshot for a file is different than our current snapshot, and is not a descendant of ours.
- overwrite: the situation when another client's current snapshot for a file is a (not necessarily direct) descendant of our current snapshot.
Overview
--------
This new design will track the history of each file using "snapshots" which are
created at each upload. Each snapshot will specify one or more parent snapshots,
forming a directed acyclic graph. A Magic-Folder user's DMD uses a flattened directory
hierarchy naming scheme, as in the original design. But, instead of pointing directly
at file contents, each file name will link to that user's latest snapshot for that file.
Inside the dmd there will also be an immutable directory containing the client's subscriptions
(read-caps to other clients' dmds).
Clients periodically poll each other's DMDs. When they see the current snapshot for a file is
different than their own current snapshot for that file, they immediately begin downloading its
contents and then walk backwards through the DAG from the new snapshot until they find their own
snapshot or a common ancestor.
For the common ancestor search to be efficient, the client will need to keep a local store (in the magic folder db) of all of the snapshots
(but not their contents) between the oldest current snapshot of any of their subscriptions and their own current snapshot.
See "local cache purging policy" below for more details.
If the new snapshot is a descendant of the client's existing snapshot, then this update
is an "overwrite" - like a git fast-forward. So, when the download of the new file completes it can overwrite
the existing local file with the new contents and update its dmd to point at the new snapshot.
If the new snapshot is not a descendant of the client's current snapshot, then the update is a
conflict. The new file is downloaded and named $filename.conflict-$user1,$user2 (including a list
of other subscriptions who have that version as their current version).
Changes to the local .conflict- file are not tracked. When that file disappears
(either by deletion, or being renamed) a new snapshot for the conflicting file is
created which has two parents - the client's snapshot prior to the conflict, and the
new conflicting snapshot. If multiple .conflict files are deleted or renamed in a short
period of time, a single conflict-resolving snapshot with more than two parents can be created.
! I think this behavior will confuse users.
Tahoe-LAFS snapshot objects
---------------------------
These Tahoe-LAFS snapshot objects only track the history of a single file, not a directory hierarchy.
Snapshot objects contain only two field types:
- ``Content``: an immutable capability of the file contents (omitted if deletion snapshot)
- ``Parent0..N``: immutable capabilities representing parent snapshots
Therefore in this system an interesting side effect of this Tahoe snapshot object is that there is no
snapshot author. The only notion of an identity in the Magic-Folder system is the write capability of the user's DMD.
The snapshot object is an immutable directory which looks like this:
content -> immutable cap to file content
parent0 -> immutable cap to a parent snapshot object
parent1..N -> more parent snapshots
Snapshot Author Identity
------------------------
Snapshot identity might become an important feature so that bad actors
can be recognized and other clients can stop "subscribing" to (polling for) updates from them.
Perhaps snapshots could be signed by the user's Magic-Folder write key for this purpose? Probably a bad idea to reuse the write-cap key for this. Better to introduce ed25519 identity keys which can (optionally) sign snapshot contents and store the signature as another member of the immutable directory.
Conflict Resolution
-------------------
detection of conflicts
``````````````````````
A Magic-Folder client updates a given file's current snapshot link to a snapshot which is a descendent
of the previous snapshot. For a given file, let's say "file1", Alice can detect that Bob's DMD has a "file1"
that links to a snapshot which conflicts. Two snapshots conflict if one is not an ancestor of the other.
a possible UI for resolving conflicts
`````````````````````````````````````
If Alice links a conflicting snapshot object for a file named "file1",
Bob and Carole will see a file in their Magic-Folder called "file1.conflicted.Alice".
Alice conversely will see an additional file called "file1.conflicted.previous".
If Alice wishes to resolve the conflict with her new version of the file then
she simply deletes the file called "file1.conflicted.previous". If she wants to
choose the other version then she moves it into place:
mv file1.conflicted.previous file1
This scheme works for N number of conflicts. Bob for instance could choose
the same resolution for the conflict, like this:
mv file1.Alice file1
Deletion propagation and eventual Garbage Collection
----------------------------------------------------
When a user deletes a file, this is represented by a link from their DMD file
object to a deletion snapshot. Eventually all users will link this deletion
snapshot into their DMD. When all users have the link then they locally cache
the deletion snapshot and remove the link to that file in their DMD.
Deletions can of course be undeleted; this means creating a new snapshot
object that specifies itself a descent of the deletion snapshot.
Clients periodically renew leases to all capabilities recursively linked
to in their DMD. Files which are unlinked by ALL the users of a
given Magic-Folder will eventually be garbage collected.
Lease expirey duration must be tuned properly by storage servers such that
Garbage Collection does not occur too frequently.
Performance Considerations
--------------------------
local changes
`````````````
Our old scheme requires two remote Tahoe-LAFS operations per local file modification:
1. upload new file contents (as an immutable file)
2. modify mutable directory (DMD) to link to the immutable file cap
Our new scheme requires three remote operations:
1. upload new file contents (as in immutable file)
2. upload immutable directory representing Tahoe-LAFS snapshot object
3. modify mutable directory (DMD) to link to the immutable snapshot object
remote changes
``````````````
Our old scheme requires one remote Tahoe-LAFS operation per remote file modification (not counting the polling of the dmd):
1. Download new file content
Our new scheme requires a minimum of two remote operations (not counting the polling of the dmd) for conflicting downloads, or three remote operations for overwrite downloads:
1. Download new snapshot object
2. Download the content it points to
3. If the download is an overwrite, modify the DMD to indicate that the downloaded version is their current version.
If the new snapshot is not a direct descendant of our current snapshot or the other party's previous snapshot we saw, we will also need to download more snapshots to determine if it is a conflict or an overwrite. However, those can be done in
parallel with the content download since we will need to download the content in either case.
While the old scheme is obviously more efficient, we think that the properties provided by the new scheme make it worth the additional cost.
Physical updates to the DMD overiouslly need to be serialized, so multiple logical updates should be combined when an update is already in progress.
conflict detection and local caching
````````````````````````````````````
Local caching of snapshots is important for performance.
We refer to the client's local snapshot cache as the ``magic-folder db``.
Conflict detection can be expensive because it may require the client
to download many snapshots from the other user's DMD in order to try
and find it's own current snapshot or a descendent. The cost of scanning
the remote DMDs should not be very high unless the client conducting the
scan has lots of history to download because of being offline for a long
time while many new snapshots were distributed.
local cache purging policy
``````````````````````````
The client's current snapshot for each file should be cached at all times.
When all clients' views of a file are synchronized (they all have the same
snapshot for that file), no ancestry for that file needs to be cached.
When clients' views of a file are *not* synchronized, the most recent
common ancestor of all clients' snapshots must be kept cached, as must
all intermediate snapshots.
Local Merge Property
--------------------
Bob can in fact, set a pre-existing directory (with files) as his new Magic-Folder directory, resulting
in a merge of the Magic-Folder with Bob's local directory. Filename collisions will result in conflicts
because Bob's new snapshots are not descendent's of the existing Magic-Folder file snapshots.
Example: simultaneous update with four parties:
1. A, B, C, D are in sync for file "foo" at snapshot X
2. A and B simultaneously change the file, creating snapshots XA and XB (both descendants of X).
3. C hears about XA first, and D hears about XB first. Both accept an overwrite.
4. All four parties hear about the other update they hadn't heard about yet.
5. Result:
- everyone's local file "foo" has the content pointed to by the snapshot in their DMD's "foo" entry
- A and C's DMDs each have the "foo" entry pointing at snapshot XA
- B and D's DMDs each have the "foo" entry pointing at snapshot XB
- A and C have a local file called foo.conflict-B,D with XB's content
- B and D have a local file called foo.conflict-A,C with XA's content
Later:
- Everyone ignores the conflict, and continue updating their local "foo". but slowly enough that there are no further conflicts, so that A and C remain in sync with eachother, and B and D remain in sync with eachother.
- A and C's foo.conflict-B,D file continues to be updated with the latest version of the file B and D are working on, and vice-versa.
- A and C edit the file at the same time again, causing a new conflict.
- Local files are now:
A: "foo", "foo.conflict-B,D", "foo.conflict-C"
C: "foo", "foo.conflict-B,D", "foo.conflict-A"
B and D: "foo", "foo.conflict-A", "foo.conflict-C"
- Finally, D decides to look at "foo.conflict-A" and "foo.conflict-C", and they manually integrate (or decide to ignore) the differences into their own local file "foo".
- D deletes their conflict files.
- D's DMD now points to a snapshot that is a descendant of everyone else's current snapshot, resolving all conflicts.
- The conflict files on A, B, and C disappear, and everyone's local file "foo" contains D's manually-merged content.
Daira: I think it is too complicated to include multiple nicknames in the .conflict files
(e.g. "foo.conflict-B,D"). It should be sufficient to have one file for each other client,
reflecting that client's latest version, regardless of who else it conflicts with.
Zooko's Design (as interpreted by Daira)
========================================
A version map is a mapping from client nickname to version number.
Definition: a version map M' strictly-follows a mapping M iff for every entry c->v
in M, there is an entry c->v' in M' such that v' > v.
Each client maintains a 'local version map' and a 'conflict version map' for each file
in its magic folder db.
If it has never written the file, then the entry for its own nickname in the local version
map is zero. The conflict version map only contains entries for nicknames B where
"$FILENAME.conflict-$B" exists.
When a client A uploads a file, it increments the version for its own nickname in its
local version map for the file, and includes that map as metadata with its upload.
A download by client A from client B is an overwrite iff the downloaded version map
strictly-follows A's local version map for that file; in this case A replaces its local
version map with the downloaded version map. Otherwise it is a conflict, and the
download is put into "$FILENAME.conflict-$B"; in this case A's
local version map remains unchanged, and the entry B->v taken from the downloaded
version map is added to its conflict version map.
If client A deletes or renames a conflict file "$FILENAME.conflict-$B", then A copies
the entry for B from its conflict version map to its local version map, deletes
the entry for B in its conflict version map, and performs another upload (with
incremented version number) of $FILENAME.
Example:
A, B, C = (10, 20, 30) everyone agrees.
A updates: (11, 20, 30)
B updates: (10, 21, 30)
C will see either A or B first. Both would be an overwrite, if considered alone.

951
docs/proposed/magic-folder/remote-to-local-sync.rst
View File

@ -1,951 +0,0 @@
Magic Folder design for remote-to-local sync
============================================
Scope
-----
In this Objective we will design remote-to-local synchronization:
* How to efficiently determine which objects (files and directories) have
to be downloaded in order to bring the current local filesystem into sync
with the newly-discovered version of the remote filesystem.
* How to distinguish overwrites, in which the remote side was aware of
your most recent version and overwrote it with a new version, from
conflicts, in which the remote side was unaware of your most recent
version when it published its new version. The latter needs to be raised
to the user as an issue the user will have to resolve and the former must
not bother the user.
* How to overwrite the (stale) local versions of those objects with the
newly acquired objects, while preserving backed-up versions of those
overwritten objects in case the user didn't want this overwrite and wants
to recover the old version.
Tickets on the Tahoe-LAFS trac with the `otf-magic-folder-objective4`_
keyword are within the scope of the remote-to-local synchronization
design.
.. _otf-magic-folder-objective4: https://tahoe-lafs.org/trac/tahoe-lafs/query?status=!closed&keywords=~otf-magic-folder-objective4
Glossary
''''''''
Object: a file or directory
DMD: distributed mutable directory
Folder: an abstract directory that is synchronized between clients.
(A folder is not the same as the directory corresponding to it on
any particular client, nor is it the same as a DMD.)
Collective: the set of clients subscribed to a given Magic Folder.
Descendant: a direct or indirect child in a directory or folder tree
Subfolder: a folder that is a descendant of a magic folder
Subpath: the path from a magic folder to one of its descendants
Write: a modification to a local filesystem object by a client
Read: a read from a local filesystem object by a client
Upload: an upload of a local object to the Tahoe-LAFS file store
Download: a download from the Tahoe-LAFS file store to a local object
Pending notification: a local filesystem change that has been detected
but not yet processed.
Representing the Magic Folder in Tahoe-LAFS
-------------------------------------------
Unlike the local case where we use inotify or ReadDirectoryChangesW to
detect filesystem changes, we have no mechanism to register a monitor for
changes to a Tahoe-LAFS directory. Therefore, we must periodically poll
for changes.
An important constraint on the solution is Tahoe-LAFS' ":doc:`write
coordination directive<../../write_coordination>`", which prohibits
concurrent writes by different storage clients to the same mutable object:
Tahoe does not provide locking of mutable files and directories. If
there is more than one simultaneous attempt to change a mutable file
or directory, then an UncoordinatedWriteError may result. This might,
in rare cases, cause the file or directory contents to be accidentally
deleted. The user is expected to ensure that there is at most one
outstanding write or update request for a given file or directory at
a time. One convenient way to accomplish this is to make a different
file or directory for each person or process that wants to write.
Since it is a goal to allow multiple users to write to a Magic Folder,
if the write coordination directive remains the same as above, then we
will not be able to implement the Magic Folder as a single Tahoe-LAFS
DMD. In general therefore, we will have multiple DMDs —spread across
clients— that together represent the Magic Folder. Each client in a
Magic Folder collective polls the other clients' DMDs in order to detect
remote changes.
Six possible designs were considered for the representation of subfolders
of the Magic Folder:
1. All subfolders written by a given Magic Folder client are collapsed
into a single client DMD, containing immutable files. The child name of
each file encodes the full subpath of that file relative to the Magic
Folder.
2. The DMD tree under a client DMD is a direct copy of the folder tree
written by that client to the Magic Folder. Not all subfolders have
corresponding DMDs; only those to which that client has written files or
child subfolders.
3. The directory tree under a client DMD is a ``tahoe backup`` structure
containing immutable snapshots of the folder tree written by that client
to the Magic Folder. As in design 2, only objects written by that client
are present.
4. *Each* client DMD contains an eventually consistent mirror of all
files and folders written by *any* Magic Folder client. Thus each client
must also copy changes made by other Magic Folder clients to its own
client DMD.
5. *Each* client DMD contains a ``tahoe backup`` structure containing
immutable snapshots of all files and folders written by *any* Magic
Folder client. Thus each client must also create another snapshot in its
own client DMD when changes are made by another client. (It can potentially
batch changes, subject to latency requirements.)
6. The write coordination problem is solved by implementing `two-phase
commit`_. Then, the representation consists of a single DMD tree which is
written by all clients.
.. _`two-phase commit`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1755
Here is a summary of advantages and disadvantages of each design:
+----------------------------+
| Key |
+=======+====================+
| \+\+ | major advantage |
+-------+--------------------+
| \+ | minor advantage |
+-------+--------------------+
| ‒ | minor disadvantage |
+-------+--------------------+
| ‒ ‒ | major disadvantage |
+-------+--------------------+
| ‒ ‒ ‒ | showstopper |
+-------+--------------------+
123456+: All designs have the property that a recursive add-lease operation
starting from a *collective directory* containing all of the client DMDs,
will find all of the files and directories used in the Magic Folder
representation. Therefore the representation is compatible with :doc:`garbage
collection <../../garbage-collection>`, even when a pre-Magic-Folder client
does the lease marking.
123456+: All designs avoid "breaking" pre-Magic-Folder clients that read
a directory or file that is part of the representation.
456++: Only these designs allow a readcap to one of the client
directories —or one of their subdirectories— to be directly shared
with other Tahoe-LAFS clients (not necessarily Magic Folder clients),
so that such a client sees all of the contents of the Magic Folder.
Note that this was not a requirement of the OTF proposal, although it
is useful.
135+: A Magic Folder client has only one mutable Tahoe-LAFS object to
monitor per other client. This minimizes communication bandwidth for
polling, or alternatively the latency possible for a given polling
bandwidth.
1236+: A client does not need to make changes to its own DMD that repeat
changes that another Magic Folder client had previously made. This reduces
write bandwidth and complexity.
1‒: If the Magic Folder has many subfolders, their files will all be
collapsed into the same DMD, which could get quite large. In practice a
single DMD can easily handle the number of files expected to be written
by a client, so this is unlikely to be a significant issue.
123‒ ‒: In these designs, the set of files in a Magic Folder is
represented as the union of the files in all client DMDs. However,
when a file is modified by more than one client, it will be linked
from multiple client DMDs. We therefore need a mechanism, such as a
version number or a monotonically increasing timestamp, to determine
which copy takes priority.
35‒ ‒: When a Magic Folder client detects a remote change, it must
traverse an immutable directory structure to see what has changed.
Completely unchanged subtrees will have the same URI, allowing some of
this traversal to be shortcutted.
24‒ ‒ ‒: When a Magic Folder client detects a remote change, it must
traverse a mutable directory structure to see what has changed. This is
more complex and less efficient than traversing an immutable structure,
because shortcutting is not possible (each DMD retains the same URI even
if a descendant object has changed), and because the structure may change
while it is being traversed. Also the traversal needs to be robust
against cycles, which can only occur in mutable structures.
45‒ ‒: When a change occurs in one Magic Folder client, it will propagate
to all the other clients. Each client will therefore see multiple
representation changes for a single logical change to the Magic Folder
contents, and must suppress the duplicates. This is particularly
problematic for design 4 where it interacts with the preceding issue.
4‒ ‒ ‒, 5‒ ‒: There is the potential for client DMDs to get "out of sync"
with each other, potentially for long periods if errors occur. Thus each
client must be able to "repair" its client directory (and its
subdirectory structure) concurrently with performing its own writes. This
is a significant complexity burden and may introduce failure modes that
could not otherwise happen.
6‒ ‒ ‒: While two-phase commit is a well-established protocol, its
application to Tahoe-LAFS requires significant design work, and may still
leave some corner cases of the write coordination problem unsolved.
+------------------------------------------------+-----------------------------------------+
| Design Property | Designs Proposed |
+================================================+======+======+======+======+======+======+
| **advantages** | *1* | *2* | *3* | *4* | *5* | *6* |
+------------------------------------------------+------+------+------+------+------+------+
| Compatible with garbage collection |\+ |\+ |\+ |\+ |\+ |\+ |
+------------------------------------------------+------+------+------+------+------+------+
| Does not break old clients |\+ |\+ |\+ |\+ |\+ |\+ |
+------------------------------------------------+------+------+------+------+------+------+
| Allows direct sharing | | | |\+\+ |\+\+ |\+\+ |
+------------------------------------------------+------+------+------+------+------+------+
| Efficient use of bandwidth |\+ | |\+ | |\+ | |
+------------------------------------------------+------+------+------+------+------+------+
| No repeated changes |\+ |\+ |\+ | | |\+ |
+------------------------------------------------+------+------+------+------+------+------+
| **disadvantages** | *1* | *2* | *3* | *4* | *5* | *6* |
+------------------------------------------------+------+------+------+------+------+------+
| Can result in large DMDs |‒ | | | | | |
+------------------------------------------------+------+------+------+------+------+------+
| Need version number to determine priority |‒ ‒ |‒ ‒ |‒ ‒ | | | |
+------------------------------------------------+------+------+------+------+------+------+
| Must traverse immutable directory structure | | |‒ ‒ | |‒ ‒ | |
+------------------------------------------------+------+------+------+------+------+------+
| Must traverse mutable directory structure | |‒ ‒ ‒ | |‒ ‒ ‒ | | |
+------------------------------------------------+------+------+------+------+------+------+
| Must suppress duplicate representation changes | | | |‒ ‒ |‒ ‒ | |
+------------------------------------------------+------+------+------+------+------+------+
| "Out of sync" problem | | | |‒ ‒ ‒ |‒ ‒ | |
+------------------------------------------------+------+------+------+------+------+------+
| Unsolved design problems | | | | | |‒ ‒ ‒ |
+------------------------------------------------+------+------+------+------+------+------+
Evaluation of designs
'''''''''''''''''''''
Designs 2 and 3 have no significant advantages over design 1, while
requiring higher polling bandwidth and greater complexity due to the need
to create subdirectories. These designs were therefore rejected.
Design 4 was rejected due to the out-of-sync problem, which is severe
and possibly unsolvable for mutable structures.
For design 5, the out-of-sync problem is still present but possibly
solvable. However, design 5 is substantially more complex, less efficient
in bandwidth/latency, and less scalable in number of clients and
subfolders than design 1. It only gains over design 1 on the ability to
share directory readcaps to the Magic Folder (or subfolders), which was
not a requirement. It would be possible to implement this feature in
future by switching to design 6.
For the time being, however, design 6 was considered out-of-scope for
this project.
Therefore, design 1 was chosen. That is:
All subfolders written by a given Magic Folder client are collapsed
into a single client DMD, containing immutable files. The child name
of each file encodes the full subpath of that file relative to the
Magic Folder.
Each directory entry in a DMD also stores a version number, so that the
latest version of a file is well-defined when it has been modified by
multiple clients.
To enable representing empty directories, a client that creates a
directory should link a corresponding zero-length file in its DMD,
at a name that ends with the encoded directory separator character.
We want to enable dynamic configuration of the membership of a Magic
Folder collective, without having to reconfigure or restart each client
when another client joins. To support this, we have a single collective
directory that links to all of the client DMDs, named by their client
nicknames. If the collective directory is mutable, then it is possible
to change its contents in order to add clients. Note that a client DMD
should not be unlinked from the collective directory unless all of its
files are first copied to some other client DMD.
A client needs to be able to write to its own DMD, and read from other DMDs.
To be consistent with the `Principle of Least Authority`_, each client's
reference to its own DMD is a write capability, whereas its reference
to the collective directory is a read capability. The latter transitively
grants read access to all of the other client DMDs and the files linked
from them, as required.
.. _`Principle of Least Authority`: http://www.eros-os.org/papers/secnotsep.pdf
Design and implementation of the user interface for maintaining this
DMD structure and configuration will be addressed in Objectives 5 and 6.
During operation, each client will poll for changes on other clients
at a predetermined frequency. On each poll, it will reread the collective
directory (to allow for added or removed clients), and then read each
client DMD linked from it.
"Hidden" files, and files with names matching the patterns used for backup,
temporary, and conflicted files, will be ignored, i.e. not synchronized
in either direction. A file is hidden if it has a filename beginning with
"." (on any platform), or has the hidden or system attribute on Windows.
Conflict Detection and Resolution
---------------------------------
The combination of local filesystems and distributed objects is
an example of shared state concurrency, which is highly error-prone
and can result in race conditions that are complex to analyze.
Unfortunately we have no option but to use shared state in this
situation.
We call the resulting design issues "dragons" (as in "Here be dragons"),
which as a convenient mnemonic we have named after the classical
Greek elements Earth, Fire, Air, and Water.
Note: all filenames used in the following sections are examples,
and the filename patterns we use in the actual implementation may
differ. The actual patterns will probably include timestamps, and
for conflicted files, the nickname of the client that last changed
the file.
Earth Dragons: Collisions between local filesystem operations and downloads
'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
Write/download collisions
~~~~~~~~~~~~~~~~~~~~~~~~~
Suppose that Alice's Magic Folder client is about to write a
version of ``foo`` that it has downloaded in response to a remote
change.
The criteria for distinguishing overwrites from conflicts are
described later in the `Fire Dragons`_ section. Suppose that the
remote change has been initially classified as an overwrite.
(As we will see, it may be reclassified in some circumstances.)
.. _`Fire Dragons`: #fire-dragons-distinguishing-conflicts-from-overwrites
Note that writing a file that does not already have an entry in the
:ref:`magic folder db<filesystem_integration-local-scanning-and-database>` is
initially classed as an overwrite.
A *write/download collision* occurs when another program writes
to ``foo`` in the local filesystem, concurrently with the new
version being written by the Magic Folder client. We need to
ensure that this does not cause data loss, as far as possible.
An important constraint on the design is that on Windows, it is
not possible to rename a file to the same name as an existing
file in that directory. Also, on Windows it may not be possible to
delete or rename a file that has been opened by another process
(depending on the sharing flags specified by that process).
Therefore we need to consider carefully how to handle failure
conditions.
In our proposed design, Alice's Magic Folder client follows
this procedure for an overwrite in response to a remote change:
1. Write a temporary file, say ``.foo.tmp``.
2. Use the procedure described in the `Fire Dragons_` section
to obtain an initial classification as an overwrite or a
conflict. (This takes as input the ``last_downloaded_uri``
field from the directory entry of the changed ``foo``.)
3. Set the ``mtime`` of the replacement file to be at least *T* seconds
before the current local time. Stat the replacement file
to obtain its ``mtime`` and ``ctime`` as stored in the local
filesystem, and update the file's last-seen statinfo in
the magic folder db with this information. (Note that the
retrieved ``mtime`` may differ from the one that was set due
to rounding.)
4. Perform a ''file replacement'' operation (explained below)
with backup filename ``foo.backup``, replaced file ``foo``,
and replacement file ``.foo.tmp``. If any step of this
operation fails, reclassify as a conflict and stop.
To reclassify as a conflict, attempt to rename ``.foo.tmp`` to
``foo.conflicted``, suppressing errors.
The implementation of file replacement differs between Unix
and Windows. On Unix, it can be implemented as follows:
* 4a. Stat the replaced path, and set the permissions of the
replacement file to be the same as the replaced file,
bitwise-or'd with octal 600 (``rw-------``). If the replaced
file does not exist, set the permissions according to the
user's umask. If there is a directory at the replaced path,
fail.
* 4b. Attempt to move the replaced file (``foo``) to the
backup filename (``foo.backup``). If an ``ENOENT`` error
occurs because the replaced file does not exist, ignore this
error and continue with steps 4c and 4d.
* 4c. Attempt to create a hard link at the replaced filename
(``foo``) pointing to the replacement file (``.foo.tmp``).
* 4d. Attempt to unlink the replacement file (``.foo.tmp``),
suppressing errors.
Note that, if there is no conflict, the entry for ``foo``
recorded in the :ref:`magic folder
db<filesystem_integration-local-scanning-and-database>` will
reflect the ``mtime`` set in step 3. The move operation in step
4b will cause a ``MOVED_FROM`` event for ``foo``, and the link
operation in step 4c will cause an ``IN_CREATE`` event for
``foo``. However, these events will not trigger an upload,
because they are guaranteed to be processed only after the file
replacement has finished, at which point the last-seen statinfo
recorded in the database entry will exactly match the metadata
for the file's inode on disk. (The two hard links — ``foo``
and, while it still exists, ``.foo.tmp`` — share the same inode
and therefore the same metadata.)
On Windows, file replacement can be implemented by a call to
the `ReplaceFileW`_ API (with the
``REPLACEFILE_IGNORE_MERGE_ERRORS`` flag). If an error occurs
because the replaced file does not exist, then we ignore this
error and attempt to move the replacement file to the replaced
file.
Similar to the Unix case, the `ReplaceFileW`_ operation will
cause one or more change notifications for ``foo``. The replaced
``foo`` has the same ``mtime`` as the replacement file, and so any
such notification(s) will not trigger an unwanted upload.
.. _`ReplaceFileW`: https://msdn.microsoft.com/en-us/library/windows/desktop/aa365512%28v=vs.85%29.aspx
To determine whether this procedure adequately protects against data
loss, we need to consider what happens if another process attempts to
update ``foo``, for example by renaming ``foo.other`` to ``foo``.
This requires us to analyze all possible interleavings between the
operations performed by the Magic Folder client and the other process.
(Note that atomic operations on a directory are totally ordered.)
The set of possible interleavings differs between Windows and Unix.
On Unix, for the case where the replaced file already exists, we have:
* Interleaving A: the other process' rename precedes our rename in
step 4b, and we get an ``IN_MOVED_TO`` event for its rename by
step 2. Then we reclassify as a conflict; its changes end up at
``foo`` and ours end up at ``foo.conflicted``. This avoids data
loss.
* Interleaving B: its rename precedes ours in step 4b, and we do
not get an event for its rename by step 2. Its changes end up at
``foo.backup``, and ours end up at ``foo`` after being linked there
in step 4c. This avoids data loss.
* Interleaving C: its rename happens between our rename in step 4b,
and our link operation in step 4c of the file replacement. The
latter fails with an ``EEXIST`` error because ``foo`` already
exists. We reclassify as a conflict; the old version ends up at
``foo.backup``, the other process' changes end up at ``foo``, and
ours at ``foo.conflicted``. This avoids data loss.
* Interleaving D: its rename happens after our link in step 4c, and
causes an ``IN_MOVED_TO`` event for ``foo``. Its rename also changes
the ``mtime`` for ``foo`` so that it is different from the ``mtime``
calculated in step 3, and therefore different from the metadata
recorded for ``foo`` in the magic folder db. (Assuming no system
clock changes, its rename will set an ``mtime`` timestamp
corresponding to a time after step 4c, which is after the timestamp
*T* seconds before step 4a, provided that *T* seconds is
sufficiently greater than the timestamp granularity.) Therefore, an
upload will be triggered for ``foo`` after its change, which is
correct and avoids data loss.
If the replaced file did not already exist, an ``ENOENT`` error
occurs at step 4b, and we continue with steps 4c and 4d. The other
process' rename races with our link operation in step 4c. If the
other process wins the race then the effect is similar to
Interleaving C, and if we win the race this it is similar to
Interleaving D. Either case avoids data loss.
On Windows, the internal implementation of `ReplaceFileW`_ is similar
to what we have described above for Unix; it works like this:
* 4a′. Copy metadata (which does not include ``mtime``) from the
replaced file (``foo``) to the replacement file (``.foo.tmp``).
* 4b′. Attempt to move the replaced file (``foo``) onto the
backup filename (``foo.backup``), deleting the latter if it
already exists.
* 4c′. Attempt to move the replacement file (``.foo.tmp``) to the
replaced filename (``foo``); fail if the destination already
exists.
Notice that this is essentially the same as the algorithm we use
for Unix, but steps 4c and 4d on Unix are combined into a single
step 4c′. (If there is a failure at steps 4c′ after step 4b′ has
completed, the `ReplaceFileW`_ call will fail with return code
``ERROR_UNABLE_TO_MOVE_REPLACEMENT_2``. However, it is still
preferable to use this API over two `MoveFileExW`_ calls, because
it retains the attributes and ACLs of ``foo`` where possible.
Also note that if the `ReplaceFileW`_ call fails with
``ERROR_FILE_NOT_FOUND`` because the replaced file does not exist,
then the replacment operation ignores this error and continues with
the equivalent of step 4c′, as on Unix.)
However, on Windows the other application will not be able to
directly rename ``foo.other`` onto ``foo`` (which would fail because
the destination already exists); it will have to rename or delete
``foo`` first. Without loss of generality, let's say ``foo`` is
deleted. This complicates the interleaving analysis, because we
have two operations done by the other process interleaving with
three done by the magic folder process (rather than one operation
interleaving with four as on Unix).
So on Windows, for the case where the replaced file already exists,
we have:
* Interleaving A′: the other process' deletion of ``foo`` and its
rename of ``foo.other`` to ``foo`` both precede our rename in
step 4b. We get an event corresponding to its rename by step 2.
Then we reclassify as a conflict; its changes end up at ``foo``
and ours end up at ``foo.conflicted``. This avoids data loss.
* Interleaving B′: the other process' deletion of ``foo`` and its
rename of ``foo.other`` to ``foo`` both precede our rename in
step 4b. We do not get an event for its rename by step 2.
Its changes end up at ``foo.backup``, and ours end up at ``foo``
after being moved there in step 4c′. This avoids data loss.
* Interleaving C′: the other process' deletion of ``foo`` precedes
our rename of ``foo`` to ``foo.backup`` done by `ReplaceFileW`_,
but its rename of ``foo.other`` to ``foo`` does not, so we get
an ``ERROR_FILE_NOT_FOUND`` error from `ReplaceFileW`_ indicating
that the replaced file does not exist. We ignore this error and
attempt to move ``foo.tmp`` to ``foo``, racing with the other
process which is attempting to move ``foo.other`` to ``foo``.
If we win the race, then our changes end up at ``foo``, and the
other process' move fails. If the other process wins the race,
then its changes end up at ``foo``, our move fails, and we
reclassify as a conflict, so that our changes end up at
``foo.conflicted``. Either possibility avoids data loss.
* Interleaving D′: the other process' deletion and/or rename happen
during the call to `ReplaceFileW`_, causing the latter to fail.
There are two subcases:
* if the error is ``ERROR_UNABLE_TO_MOVE_REPLACEMENT_2``, then
``foo`` is renamed to ``foo.backup`` and ``.foo.tmp`` remains
at its original name after the call.
* for all other errors, ``foo`` and ``.foo.tmp`` both remain at
their original names after the call.
In both subcases, we reclassify as a conflict and rename ``.foo.tmp``
to ``foo.conflicted``. This avoids data loss.
* Interleaving E′: the other process' deletion of ``foo`` and attempt
to rename ``foo.other`` to ``foo`` both happen after all internal
operations of `ReplaceFileW`_ have completed. This causes deletion
and rename events for ``foo`` (which will in practice be merged due
to the pending delay, although we don't rely on that for
correctness). The rename also changes the ``mtime`` for ``foo`` so
that it is different from the ``mtime`` calculated in step 3, and
therefore different from the metadata recorded for ``foo`` in the
magic folder db. (Assuming no system clock changes, its rename will
set an ``mtime`` timestamp corresponding to a time after the
internal operations of `ReplaceFileW`_ have completed, which is
after the timestamp *T* seconds before `ReplaceFileW`_ is called,
provided that *T* seconds is sufficiently greater than the timestamp
granularity.) Therefore, an upload will be triggered for ``foo``
after its change, which is correct and avoids data loss.
.. _`MoveFileExW`: https://msdn.microsoft.com/en-us/library/windows/desktop/aa365240%28v=vs.85%29.aspx
If the replaced file did not already exist, we get an
``ERROR_FILE_NOT_FOUND`` error from `ReplaceFileW`_, and attempt to
move ``foo.tmp`` to ``foo``. This is similar to Interleaving C, and
either possibility for the resulting race avoids data loss.
We also need to consider what happens if another process opens ``foo``
and writes to it directly, rather than renaming another file onto it:
* On Unix, open file handles refer to inodes, not paths. If the other
process opens ``foo`` before it has been renamed to ``foo.backup``,
and then closes the file, changes will have been written to the file
at the same inode, even if that inode is now linked at ``foo.backup``.
This avoids data loss.
* On Windows, we have two subcases, depending on whether the sharing
flags specified by the other process when it opened its file handle
included ``FILE_SHARE_DELETE``. (This flag covers both deletion and
rename operations.)
i. If the sharing flags *do not* allow deletion/renaming, the
`ReplaceFileW`_ operation will fail without renaming ``foo``.
In this case we will end up with ``foo`` changed by the other
process, and the downloaded file still in ``foo.tmp``.
This avoids data loss.
ii. If the sharing flags *do* allow deletion/renaming, then
data loss or corruption may occur. This is unavoidable and
can be attributed to other process making a poor choice of
sharing flags (either explicitly if it used `CreateFile`_, or
via whichever higher-level API it used).
.. _`CreateFile`: https://msdn.microsoft.com/en-us/library/windows/desktop/aa363858%28v=vs.85%29.aspx
Note that it is possible that another process tries to open the file
between steps 4b and 4c (or 4b′ and 4c′ on Windows). In this case the
open will fail because ``foo`` does not exist. Nevertheless, no data
will be lost, and in many cases the user will be able to retry the
operation.
Above we only described the case where the download was initially
classified as an overwrite. If it was classed as a conflict, the
procedure is the same except that we choose a unique filename
for the conflicted file (say, ``foo.conflicted_unique``). We write
the new contents to ``.foo.tmp`` and then rename it to
``foo.conflicted_unique`` in such a way that the rename will fail
if the destination already exists. (On Windows this is a simple
rename; on Unix it can be implemented as a link operation followed
by an unlink, similar to steps 4c and 4d above.) If this fails
because another process wrote ``foo.conflicted_unique`` after we
chose the filename, then we retry with a different filename.
Read/download collisions
~~~~~~~~~~~~~~~~~~~~~~~~
A *read/download collision* occurs when another program reads
from ``foo`` in the local filesystem, concurrently with the new
version being written by the Magic Folder client. We want to
ensure that any successful attempt to read the file by the other
program obtains a consistent view of its contents.
On Unix, the above procedure for writing downloads is sufficient
to achieve this. There are three cases:
* A. The other process opens ``foo`` for reading before it is
renamed to ``foo.backup``. Then the file handle will continue to
refer to the old file across the rename, and the other process
will read the old contents.
* B. The other process attempts to open ``foo`` after it has been
renamed to ``foo.backup``, and before it is linked in step c.
The open call fails, which is acceptable.
* C. The other process opens ``foo`` after it has been linked to
the new file. Then it will read the new contents.
On Windows, the analysis is very similar, but case A′ needs to
be split into two subcases, depending on the sharing mode the other
process uses when opening the file for reading:
* A′. The other process opens ``foo`` before the Magic Folder
client's attempt to rename ``foo`` to ``foo.backup`` (as part
of the implementation of `ReplaceFileW`_). The subcases are:
i. The other process uses sharing flags that deny deletion and
renames. The `ReplaceFileW`_ call fails, and the download is
reclassified as a conflict. The downloaded file ends up at
``foo.conflicted``, which is correct.
ii. The other process uses sharing flags that allow deletion
and renames. The `ReplaceFileW`_ call succeeds, and the
other process reads inconsistent data. This can be attributed
to a poor choice of sharing flags by the other process.
* B′. The other process attempts to open ``foo`` at the point
during the `ReplaceFileW`_ call where it does not exist.
The open call fails, which is acceptable.
* C′. The other process opens ``foo`` after it has been linked to
the new file. Then it will read the new contents.
For both write/download and read/download collisions, we have
considered only interleavings with a single other process, and
only the most common possibilities for the other process'
interaction with the file. If multiple other processes are
involved, or if a process performs operations other than those
considered, then we cannot say much about the outcome in general;
however, we believe that such cases will be much less common.
Fire Dragons: Distinguishing conflicts from overwrites
''''''''''''''''''''''''''''''''''''''''''''''''''''''
When synchronizing a file that has changed remotely, the Magic Folder
client needs to distinguish between overwrites, in which the remote
side was aware of your most recent version (if any) and overwrote it
with a new version, and conflicts, in which the remote side was unaware
of your most recent version when it published its new version. Those two
cases have to be handled differently — the latter needs to be raised
to the user as an issue the user will have to resolve and the former
must not bother the user.
For example, suppose that Alice's Magic Folder client sees a change
to ``foo`` in Bob's DMD. If the version it downloads from Bob's DMD
is "based on" the version currently in Alice's local filesystem at
the time Alice's client attempts to write the downloaded file ‒or if
there is no existing version in Alice's local filesystem at that time‒
then it is an overwrite. Otherwise it is initially classified as a
conflict.
This initial classification is used by the procedure for writing a
file described in the `Earth Dragons`_ section above. As explained
in that section, we may reclassify an overwrite as a conflict if an
error occurs during the write procedure.
.. _`Earth Dragons`: #earth-dragons-collisions-between-local-filesystem-operations-and-downloads
In order to implement this policy, we need to specify how the
"based on" relation between file versions is recorded and updated.
We propose to record this information:
* in the :ref:`magic folder
db<filesystem_integration-local-scanning-and-database>`, for
local files;
* in the Tahoe-LAFS directory metadata, for files stored in the
Magic Folder.
In the magic folder db we will add a *last-downloaded record*,
consisting of ``last_downloaded_uri`` and ``last_downloaded_timestamp``
fields, for each path stored in the database. Whenever a Magic Folder
client downloads a file, it stores the downloaded version's URI and
the current local timestamp in this record. Since only immutable
files are used, the URI will be an immutable file URI, which is
deterministically and uniquely derived from the file contents and
the Tahoe-LAFS node's :doc:`convergence secret<../../convergence-secret>`.
(Note that the last-downloaded record is updated regardless of
whether the download is an overwrite or a conflict. The rationale
for this to avoid "conflict loops" between clients, where every
new version after the first conflict would be considered as another
conflict.)
Later, in response to a local filesystem change at a given path, the
Magic Folder client reads the last-downloaded record associated with
that path (if any) from the database and then uploads the current
file. When it links the uploaded file into its client DMD, it
includes the ``last_downloaded_uri`` field in the metadata of the
directory entry, overwriting any existing field of that name. If
there was no last-downloaded record associated with the path, this
field is omitted.
Note that ``last_downloaded_uri`` field does *not* record the URI of
the uploaded file (which would be redundant); it records the URI of
the last download before the local change that caused the upload.
The field will be absent if the file has never been downloaded by
this client (i.e. if it was created on this client and no change
by any other client has been detected).
A possible refinement also takes into account the
``last_downloaded_timestamp`` field from the magic folder db, and
compares it to the timestamp of the change that caused the upload
(which should be later, assuming no system clock changes).
If the duration between these timestamps is very short, then we
are uncertain about whether the process on Bob's system that wrote
the local file could have taken into account the last download.
We can use this information to be conservative about treating
changes as conflicts. So, if the duration is less than a configured
threshold, we omit the ``last_downloaded_uri`` field from the
metadata. This will have the effect of making other clients treat
this change as a conflict whenever they already have a copy of the
file.
Conflict/overwrite decision algorithm
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Now we are ready to describe the algorithm for determining whether a
download for the file ``foo`` is an overwrite or a conflict (refining
step 2 of the procedure from the `Earth Dragons`_ section).
Let ``last_downloaded_uri`` be the field of that name obtained from
the directory entry metadata for ``foo`` in Bob's DMD (this field
may be absent). Then the algorithm is:
* 2a. Attempt to "stat" ``foo`` to get its *current statinfo* (size
in bytes, ``mtime``, and ``ctime``). If Alice has no local copy
of ``foo``, classify as an overwrite.
* 2b. Read the following information for the path ``foo`` from the
local magic folder db:
* the *last-seen statinfo*, if any (this is the size in
bytes, ``mtime``, and ``ctime`` stored in the ``local_files``
table when the file was last uploaded);
* the ``last_uploaded_uri`` field of the ``local_files`` table
for this file, which is the URI under which the file was last
uploaded.
* 2c. If any of the following are true, then classify as a conflict:
* i. there are pending notifications of changes to ``foo``;
* ii. the last-seen statinfo is either absent (i.e. there is
no entry in the database for this path), or different from the
current statinfo;
* iii. either ``last_downloaded_uri`` or ``last_uploaded_uri``
(or both) are absent, or they are different.
Otherwise, classify as an overwrite.
Air Dragons: Collisions between local writes and uploads
''''''''''''''''''''''''''''''''''''''''''''''''''''''''
Short of filesystem-specific features on Unix or the `shadow copy service`_
on Windows (which is per-volume and therefore difficult to use in this
context), there is no way to *read* the whole contents of a file
atomically. Therefore, when we read a file in order to upload it, we
may read an inconsistent version if it was also being written locally.
.. _`shadow copy service`: https://technet.microsoft.com/en-us/library/ee923636%28v=ws.10%29.aspx
A well-behaved application can avoid this problem for its writes:
* On Unix, if another process modifies a file by renaming a temporary
file onto it, then we will consistently read either the old contents
or the new contents.
* On Windows, if the other process uses sharing flags to deny reads
while it is writing a file, then we will consistently read either
the old contents or the new contents, unless a sharing error occurs.
In the case of a sharing error we should retry later, up to a
maximum number of retries.
In the case of a not-so-well-behaved application writing to a file
at the same time we read from it, the magic folder will still be
eventually consistent, but inconsistent versions may be visible to
other users' clients.
In Objective 2 we implemented a delay, called the *pending delay*,
after the notification of a filesystem change and before the file is
read in order to upload it (Tahoe-LAFS ticket `#1440`_). If another
change notification occurs within the pending delay time, the delay
is restarted. This helps to some extent because it means that if
files are written more quickly than the pending delay and less
frequently than the pending delay, we shouldn't encounter this
inconsistency.
.. _`#1440`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1440
The likelihood of inconsistency could be further reduced, even for
writes by not-so-well-behaved applications, by delaying the actual
upload for a further period —called the *stability delay*— after the
file has finished being read. If a notification occurs between the
end of the pending delay and the end of the stability delay, then
the read would be aborted and the notification requeued.
This would have the effect of ensuring that no write notifications
have been received for the file during a time window that brackets
the period when it was being read, with margin before and after
this period defined by the pending and stability delays. The delays
are intended to account for asynchronous notification of events, and
caching in the filesystem.
Note however that we cannot guarantee that the delays will be long
enough to prevent inconsistency in any particular case. Also, the
stability delay would potentially affect performance significantly
because (unlike the pending delay) it is not overlapped when there
are multiple files on the upload queue. This performance impact
could be mitigated by uploading files in parallel where possible
(Tahoe-LAFS ticket `#1459`_).
We have not yet decided whether to implement the stability delay, and
it is not planned to be implemented for the OTF objective 4 milestone.
Ticket `#2431`_ has been opened to track this idea.
.. _`#1459`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1459
.. _`#2431`: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/2431
Note that the situation of both a local process and the Magic Folder
client reading a file at the same time cannot cause any inconsistency.
Water Dragons: Handling deletion and renames
''''''''''''''''''''''''''''''''''''''''''''
Deletion of a file
~~~~~~~~~~~~~~~~~~
When a file is deleted from the filesystem of a Magic Folder client,
the most intuitive behavior is for it also to be deleted under that
name from other clients. To avoid data loss, the other clients should
actually rename their copies to a backup filename.
It would not be sufficient for a Magic Folder client that deletes
a file to implement this simply by removing the directory entry from
its DMD. Indeed, the entry may not exist in the client's DMD if it
has never previously changed the file.
Instead, the client links a zero-length file into its DMD and sets
``deleted: true`` in the directory entry metadata. Other clients
take this as a signal to rename their copies to the backup filename.
Note that the entry for this zero-length file has a version number as
usual, and later versions may restore the file.
When the downloader deletes a file (or renames it to a filename
ending in ``.backup``) in response to a remote change, a local
filesystem notification will occur, and we must make sure that this
is not treated as a local change. To do this we have the downloader
set the ``size`` field in the magic folder db to ``None`` (SQL NULL)
just before deleting the file, and suppress notifications for which
the local file does not exist, and the recorded ``size`` field is
``None``.
When a Magic Folder client restarts, we can detect files that had
been downloaded but were deleted while it was not running, because
their paths will have last-downloaded records in the magic folder db
with a ``size`` other than ``None``, and without any corresponding
local file.
Deletion of a directory
~~~~~~~~~~~~~~~~~~~~~~~
Local filesystems (unlike a Tahoe-LAFS filesystem) normally cannot
unlink a directory that has any remaining children. Therefore a
Magic Folder client cannot delete local copies of directories in
general, because they will typically contain backup files. This must
be done manually on each client if desired.
Nevertheless, a Magic Folder client that deletes a directory should
set ``deleted: true`` on the metadata entry for the corresponding
zero-length file. This avoids the directory being recreated after
it has been manually deleted from a client.
Renaming
~~~~~~~~
It is sufficient to handle renaming of a file by treating it as a
deletion and an addition under the new name.
This also applies to directories, although users may find the
resulting behavior unintuitive: all of the files under the old name
will be renamed to backup filenames, and a new directory structure
created under the new name. We believe this is the best that can be
done without imposing unreasonable implementation complexity.
Summary
-------
This completes the design of remote-to-local synchronization.
We realize that it may seem very complicated. Anecdotally, proprietary
filesystem synchronization designs we are aware of, such as Dropbox,
are said to incur similar or greater design complexity.

205
docs/proposed/magic-folder/user-interface-design.rst
View File

@ -1,205 +0,0 @@
Magic Folder user interface design
==================================
Scope
-----
In this Objective we will design a user interface to allow users to conveniently
and securely indicate which folders on some devices should be "magically" linked
to which folders on other devices.
This is a critical usability and security issue for which there is no known perfect
solution, but which we believe is amenable to a "good enough" trade-off solution.
This document explains the design and justifies its trade-offs in terms of security,
usability, and time-to-market.
Tickets on the Tahoe-LAFS trac with the `otf-magic-folder-objective6`_
keyword are within the scope of the user interface design.
.. _otf-magic-folder-objective6: https://tahoe-lafs.org/trac/tahoe-lafs/query?status=!closed&keywords=~otf-magic-folder-objective6
Glossary
''''''''
Object: a file or directory
DMD: distributed mutable directory
Folder: an abstract directory that is synchronized between clients.
(A folder is not the same as the directory corresponding to it on
any particular client, nor is it the same as a DMD.)
Collective: the set of clients subscribed to a given Magic Folder.
Diminishing: the process of deriving, from an existing capability,
another capability that gives less authority (for example, deriving a
read cap from a read/write cap).
Design Constraints
------------------
The design of the Tahoe-side representation of a Magic Folder, and the
polling mechanism that the Magic Folder clients will use to detect remote
changes was discussed in :doc:`remote-to-local-sync<remote-to-local-sync>`,
and we will not revisit that here. The assumption made by that design was
that each client would be configured with the following information:
* a write cap to its own *client DMD*.
* a read cap to a *collective directory*.
The collective directory contains links to each client DMD named by the
corresponding client's nickname.
This design was chosen to allow straightforward addition of clients without
requiring each existing client to change its configuration.
Note that each client in a Magic Folder collective has the authority to add,
modify or delete any object within the Magic Folder. It is also able to control
to some extent whether its writes will be treated by another client as overwrites
or as conflicts. However, there is still a reliability benefit to preventing a
client from accidentally modifying another client's DMD, or from accidentally
modifying the collective directory in a way that would lose data. This motivates
ensuring that each client only has access to the caps above, rather than, say,
every client having a write cap to the collective directory.
Another important design constraint is that we cannot violate the :doc:`write
coordination directive<../../write_coordination>`; that is, we cannot write to
the same mutable directory from multiple clients, even during the setup phase
when adding a client.
Within these constraints, for usability we want to minimize the number of steps
required to configure a Magic Folder collective.
Proposed Design
---------------
Three ``tahoe`` subcommands are added::
tahoe magic-folder create MAGIC: [MY_NICKNAME LOCAL_DIR]
Create an empty Magic Folder. The MAGIC: local alias is set
to a write cap which can be used to refer to this Magic Folder
in future ``tahoe magic-folder invite`` commands.
If MY_NICKNAME and LOCAL_DIR are given, the current client
immediately joins the newly created Magic Folder with that
nickname and local directory.
tahoe magic-folder invite MAGIC: THEIR_NICKNAME
Print an "invitation" that can be used to invite another
client to join a Magic Folder, with the given nickname.
The invitation must be sent to the user of the other client
over a secure channel (e.g. PGP email, OTR, or ssh).
This command will normally be run by the same client that
created the Magic Folder. However, it may be run by a
different client if the ``MAGIC:`` alias is copied to
the ``private/aliases`` file of that other client, or if
``MAGIC:`` is replaced by the write cap to which it points.
tahoe magic-folder join INVITATION LOCAL_DIR
Accept an invitation created by ``tahoe magic-folder invite``.
The current client joins the specified Magic Folder, which will
appear in the local filesystem at the given directory.
There are no commands to remove a client or to revoke an
invitation, although those are possible features that could
be added in future. (When removing a client, it is necessary
to copy each file it added to some other client's DMD, if it
is the most recent version of that file.)
Implementation
''''''''''''''
For "``tahoe magic-folder create MAGIC: [MY_NICKNAME LOCAL_DIR]``" :
1. Run "``tahoe create-alias MAGIC:``".
2. If ``MY_NICKNAME`` and ``LOCAL_DIR`` are given, do the equivalent of::
INVITATION=`tahoe invite-magic-folder MAGIC: MY_NICKNAME`
tahoe join-magic-folder INVITATION LOCAL_DIR
For "``tahoe magic-folder invite COLLECTIVE_WRITECAP NICKNAME``" :
(``COLLECTIVE_WRITECAP`` can, as a special case, be an alias such as ``MAGIC:``.)
1. Create an empty client DMD. Let its write URI be ``CLIENT_WRITECAP``.
2. Diminish ``CLIENT_WRITECAP`` to ``CLIENT_READCAP``, and
diminish ``COLLECTIVE_WRITECAP`` to ``COLLECTIVE_READCAP``.
3. Run "``tahoe ln CLIENT_READCAP COLLECTIVE_WRITECAP/NICKNAME``".
4. Print "``COLLECTIVE_READCAP+CLIENT_WRITECAP``" as the invitation,
accompanied by instructions on how to accept the invitation and
the need to send it over a secure channel.
For "``tahoe magic-folder join INVITATION LOCAL_DIR``" :
1. Parse ``INVITATION`` as ``COLLECTIVE_READCAP+CLIENT_WRITECAP``.
2. Write ``CLIENT_WRITECAP`` to the file ``magic_folder_dircap``
under the client's ``private`` directory.
3. Write ``COLLECTIVE_READCAP`` to the file ``collective_dircap``
under the client's ``private`` directory.
4. Edit the client's ``tahoe.cfg`` to set
``[magic_folder] enabled = True`` and
``[magic_folder] local.directory = LOCAL_DIR``.
Discussion
----------
The proposed design has a minor violation of the
`Principle of Least Authority`_ in order to reduce the number
of steps needed. The invoker of "``tahoe magic-folder invite``"
creates the client DMD on behalf of the invited client, and
could retain its write cap (which is part of the invitation).
.. _`Principle of Least Authority`: http://www.eros-os.org/papers/secnotsep.pdf
A possible alternative design would be for the invited client
to create its own client DMD, and send it back to the inviter
to be linked into the collective directory. However this would
require another secure communication and another command
invocation per client. Given that, as mentioned earlier, each
client in a Magic Folder collective already has the authority
to add, modify or delete any object within the Magic Folder,
we considered the potential security/reliability improvement
here not to be worth the loss of usability.
We also considered a design where each client had write access to
the collective directory. This would arguably be a more serious
violation of the Principle of Least Authority than the one above
(because all clients would have excess authority rather than just
the inviter). In any case, it was not clear how to make such a
design satisfy the :doc:`write coordination
directive<../../write_coordination>`, because the collective
directory would have needed to be written to by multiple clients.
The reliance on a secure channel to send the invitation to its
intended recipient is not ideal, since it may involve additional
software such as clients for PGP, OTR, ssh etc. However, we believe
that this complexity is necessary rather than incidental, because
there must be some way to distinguish the intended recipient from
potential attackers who would try to become members of the Magic
Folder collective without authorization. By making use of existing
channels that have likely already been set up by security-conscious
users, we avoid reinventing the wheel or imposing substantial extra
implementation costs.
The length of an invitation will be approximately the combined
length of a Tahoe-LAFS read cap and write cap. This is several
lines long, but still short enough to be cut-and-pasted successfully
if care is taken. Errors in copying the invitation can be detected
since Tahoe-LAFS cap URIs are self-authenticating.
The implementation of the ``tahoe`` subcommands is straightforward
and raises no further difficult design issues.

6
docs/windows.rst
View File

@ -77,9 +77,9 @@ If you're planning to hack on the source code, you might want to add
Dependencies
------------
Tahoe-LAFS depends upon several packages that use compiled C code, such as
zfec, pycryptopp, and others. This code must be built separately for each
platform (Windows, OS-X, and different flavors of Linux).
Tahoe-LAFS depends upon several packages that use compiled C code
(such as zfec). This code must be built separately for each platform
(Windows, OS-X, and different flavors of Linux).
Pre-compiled "wheels" of all Tahoe's dependencies are hosted on the
tahoe-lafs.org website in the ``deps/`` directory. The ``--find-links=``

186
integration/conftest.py
View File

@ -3,7 +3,7 @@ from __future__ import print_function
import sys
import shutil
from time import sleep
from os import mkdir, listdir
from os import mkdir, listdir, environ
from os.path import join, exists
from tempfile import mkdtemp, mktemp
from functools import partial
@ -15,6 +15,7 @@ from eliot import (
)
from twisted.python.procutils import which
from twisted.internet.defer import DeferredList
from twisted.internet.error import (
ProcessExitedAlready,
ProcessTerminated,
@ -30,7 +31,10 @@ from util import (
_ProcessExitedProtocol,
_create_node,
_run_node,
_cleanup_twistd_process,
_cleanup_tahoe_process,
_tahoe_runner_optional_coverage,
await_client_ready,
TahoeProcess,
)
@ -41,6 +45,10 @@ def pytest_addoption(parser):
"--keep-tempdir", action="store_true", dest="keep",
help="Keep the tmpdir with the client directories (introducer, etc)",
)
parser.addoption(
"--coverage", action="store_true", dest="coverage",
help="Collect coverage statistics",
)
@pytest.fixture(autouse=True, scope='session')
def eliot_logging():
@ -125,7 +133,7 @@ def flog_gatherer(reactor, temp_dir, flog_binary, request):
pytest_twisted.blockon(twistd_protocol.magic_seen)
def cleanup():
_cleanup_twistd_process(twistd_process, twistd_protocol.exited)
_cleanup_tahoe_process(twistd_process, twistd_protocol.exited)
flog_file = mktemp('.flog_dump')
flog_protocol = _DumpOutputProtocol(open(flog_file, 'w'))
@ -174,11 +182,11 @@ log_gatherer.furl = {log_furl}
if not exists(intro_dir):
mkdir(intro_dir)
done_proto = _ProcessExitedProtocol()
reactor.spawnProcess(
_tahoe_runner_optional_coverage(
done_proto,
sys.executable,
reactor,
request,
(
sys.executable, '-m', 'allmydata.scripts.runner',
'create-introducer',
'--listen=tcp',
'--hostname=localhost',
@ -195,19 +203,19 @@ log_gatherer.furl = {log_furl}
# but on linux it means daemonize. "tahoe run" is consistent
# between platforms.
protocol = _MagicTextProtocol('introducer running')
process = reactor.spawnProcess(
transport = _tahoe_runner_optional_coverage(
protocol,
sys.executable,
reactor,
request,
(
sys.executable, '-m', 'allmydata.scripts.runner',
'run',
intro_dir,
),
)
request.addfinalizer(partial(_cleanup_twistd_process, process, protocol.exited))
request.addfinalizer(partial(_cleanup_tahoe_process, transport, protocol.exited))
pytest_twisted.blockon(protocol.magic_seen)
return process
return TahoeProcess(transport, intro_dir)
@pytest.fixture(scope='session')
@ -241,11 +249,11 @@ log_gatherer.furl = {log_furl}
if not exists(intro_dir):
mkdir(intro_dir)
done_proto = _ProcessExitedProtocol()
reactor.spawnProcess(
_tahoe_runner_optional_coverage(
done_proto,
sys.executable,
reactor,
request,
(
sys.executable, '-m', 'allmydata.scripts.runner',
'create-introducer',
'--tor-control-port', 'tcp:localhost:8010',
'--listen=tor',
@ -262,11 +270,11 @@ log_gatherer.furl = {log_furl}
# but on linux it means daemonize. "tahoe run" is consistent
# between platforms.
protocol = _MagicTextProtocol('introducer running')
process = reactor.spawnProcess(
transport = _tahoe_runner_optional_coverage(
protocol,
sys.executable,
reactor,
request,
(
sys.executable, '-m', 'allmydata.scripts.runner',
'run',
intro_dir,
),
@ -274,14 +282,14 @@ log_gatherer.furl = {log_furl}
def cleanup():
try:
process.signalProcess('TERM')
transport.signalProcess('TERM')
pytest_twisted.blockon(protocol.exited)
except ProcessExitedAlready:
pass
request.addfinalizer(cleanup)
pytest_twisted.blockon(protocol.magic_seen)
return process
return transport
@pytest.fixture(scope='session')
@ -301,31 +309,29 @@ def tor_introducer_furl(tor_introducer, temp_dir):
include_result=False,
)
def storage_nodes(reactor, temp_dir, introducer, introducer_furl, flog_gatherer, request):
nodes = []
nodes_d = []
# start all 5 nodes in parallel
for x in range(5):
name = 'node{}'.format(x)
# tub_port = 9900 + x
nodes.append(
pytest_twisted.blockon(
_create_node(
reactor, request, temp_dir, introducer_furl, flog_gatherer, name,
web_port=None, storage=True,
)
web_port= 9990 + x
nodes_d.append(
_create_node(
reactor, request, temp_dir, introducer_furl, flog_gatherer, name,
web_port="tcp:{}:interface=localhost".format(web_port),
storage=True,
)
)
#nodes = pytest_twisted.blockon(DeferredList(nodes))
nodes_status = pytest_twisted.blockon(DeferredList(nodes_d))
nodes = []
for ok, process in nodes_status:
assert ok, "Storage node creation failed: {}".format(process)
nodes.append(process)
return nodes
@pytest.fixture(scope='session')
@log_call(action_type=u"integration:alice", include_args=[], include_result=False)
def alice(reactor, temp_dir, introducer_furl, flog_gatherer, storage_nodes, request):
try:
mkdir(join(temp_dir, 'magic-alice'))
except OSError:
pass
process = pytest_twisted.blockon(
_create_node(
reactor, request, temp_dir, introducer_furl, flog_gatherer, "alice",
@ -333,17 +339,13 @@ def alice(reactor, temp_dir, introducer_furl, flog_gatherer, storage_nodes, requ
storage=False,
)
)
await_client_ready(process)
return process
@pytest.fixture(scope='session')
@log_call(action_type=u"integration:bob", include_args=[], include_result=False)
def bob(reactor, temp_dir, introducer_furl, flog_gatherer, storage_nodes, request):
try:
mkdir(join(temp_dir, 'magic-bob'))
except OSError:
pass
process = pytest_twisted.blockon(
_create_node(
reactor, request, temp_dir, introducer_furl, flog_gatherer, "bob",
@ -351,99 +353,10 @@ def bob(reactor, temp_dir, introducer_furl, flog_gatherer, storage_nodes, reques
storage=False,
)
)
await_client_ready(process)
return process
@pytest.fixture(scope='session')
@log_call(action_type=u"integration:alice:invite", include_args=["temp_dir"])
def alice_invite(reactor, alice, temp_dir, request):
node_dir = join(temp_dir, 'alice')
with start_action(action_type=u"integration:alice:magic_folder:create"):
# FIXME XXX by the time we see "client running" in the logs, the
# storage servers aren't "really" ready to roll yet (uploads fairly
# consistently fail if we don't hack in this pause...)
import time ; time.sleep(5)
proto = _CollectOutputProtocol()
reactor.spawnProcess(
proto,
sys.executable,
[
sys.executable, '-m', 'allmydata.scripts.runner',
'magic-folder', 'create',
'--poll-interval', '2',
'--basedir', node_dir, 'magik:', 'alice',
join(temp_dir, 'magic-alice'),
]
)
pytest_twisted.blockon(proto.done)
with start_action(action_type=u"integration:alice:magic_folder:invite") as a:
proto = _CollectOutputProtocol()
reactor.spawnProcess(
proto,
sys.executable,
[
sys.executable, '-m', 'allmydata.scripts.runner',
'magic-folder', 'invite',
'--basedir', node_dir, 'magik:', 'bob',
]
)
pytest_twisted.blockon(proto.done)
invite = proto.output.getvalue()
a.add_success_fields(invite=invite)
with start_action(action_type=u"integration:alice:magic_folder:restart"):
# before magic-folder works, we have to stop and restart (this is
# crappy for the tests -- can we fix it in magic-folder?)
try:
alice.signalProcess('TERM')
pytest_twisted.blockon(alice.exited)
except ProcessExitedAlready:
pass
with start_action(action_type=u"integration:alice:magic_folder:magic-text"):
magic_text = 'Completed initial Magic Folder scan successfully'
pytest_twisted.blockon(_run_node(reactor, node_dir, request, magic_text))
return invite
@pytest.fixture(scope='session')
@log_call(
action_type=u"integration:magic_folder",
include_args=["alice_invite", "temp_dir"],
)
def magic_folder(reactor, alice_invite, alice, bob, temp_dir, request):
print("pairing magic-folder")
bob_dir = join(temp_dir, 'bob')
proto = _CollectOutputProtocol()
reactor.spawnProcess(
proto,
sys.executable,
[
sys.executable, '-m', 'allmydata.scripts.runner',
'magic-folder', 'join',
'--poll-interval', '2',
'--basedir', bob_dir,
alice_invite,
join(temp_dir, 'magic-bob'),
]
)
pytest_twisted.blockon(proto.done)
# before magic-folder works, we have to stop and restart (this is
# crappy for the tests -- can we fix it in magic-folder?)
try:
print("Sending TERM to Bob")
bob.signalProcess('TERM')
pytest_twisted.blockon(bob.exited)
except ProcessExitedAlready:
pass
magic_text = 'Completed initial Magic Folder scan successfully'
pytest_twisted.blockon(_run_node(reactor, bob_dir, request, magic_text))
return (join(temp_dir, 'magic-alice'), join(temp_dir, 'magic-bob'))
@pytest.fixture(scope='session')
def chutney(reactor, temp_dir):
chutney_dir = join(temp_dir, 'chutney')
@ -462,12 +375,13 @@ def chutney(reactor, temp_dir):
proto = _DumpOutputProtocol(None)
reactor.spawnProcess(
proto,
'/usr/bin/git',
'git',
(
'/usr/bin/git', 'clone', '--depth=1',
'git', 'clone', '--depth=1',
'https://git.torproject.org/chutney.git',
chutney_dir,
)
),
env=environ,
)
pytest_twisted.blockon(proto.done)
return chutney_dir
@ -483,6 +397,8 @@ def tor_network(reactor, temp_dir, chutney, request):
# ./chutney configure networks/basic
# ./chutney start networks/basic
env = environ.copy()
env.update({"PYTHONPATH": join(chutney_dir, "lib")})
proto = _DumpOutputProtocol(None)
reactor.spawnProcess(
proto,
@ -492,7 +408,7 @@ def tor_network(reactor, temp_dir, chutney, request):
join(chutney_dir, 'networks', 'basic'),
),
path=join(chutney_dir),
env={"PYTHONPATH": join(chutney_dir, "lib")},
env=env,
)
pytest_twisted.blockon(proto.done)
@ -505,7 +421,7 @@ def tor_network(reactor, temp_dir, chutney, request):
join(chutney_dir, 'networks', 'basic'),
),
path=join(chutney_dir),
env={"PYTHONPATH": join(chutney_dir, "lib")},
env=env,
)
pytest_twisted.blockon(proto.done)
@ -519,7 +435,7 @@ def tor_network(reactor, temp_dir, chutney, request):
join(chutney_dir, 'networks', 'basic'),
),
path=join(chutney_dir),
env={"PYTHONPATH": join(chutney_dir, "lib")},
env=env,
)
try:
pytest_twisted.blockon(proto.done)
@ -538,7 +454,7 @@ def tor_network(reactor, temp_dir, chutney, request):
join(chutney_dir, 'networks', 'basic'),
),
path=join(chutney_dir),
env={"PYTHONPATH": join(chutney_dir, "lib")},
env=env,
)
pytest_twisted.blockon(proto.done)
request.addfinalizer(cleanup)

4
integration/test_aaa_aardvark.py
View File

@ -16,7 +16,3 @@ def test_create_introducer(introducer):
def test_create_storage(storage_nodes):
print("Created {} storage nodes".format(len(storage_nodes)))
def test_create_alice_bob_magicfolder(magic_folder):
print("Alice and Bob have paired magic-folders")

461
integration/test_magic_folder.py
View File

@ -1,461 +0,0 @@
import sys
import time
import shutil
from os import mkdir, unlink, utime
from os.path import join, exists, getmtime
import util
import pytest_twisted
# tests converted from check_magicfolder_smoke.py
# see "conftest.py" for the fixtures (e.g. "magic_folder")
def test_eliot_logs_are_written(alice, bob, temp_dir):
# The integration test configuration arranges for this logging
# configuration. Verify it actually does what we want.
#
# The alice and bob arguments looks unused but they actually tell pytest
# to set up all the magic-folder stuff. The assertions here are about
# side-effects of that setup.
assert exists(join(temp_dir, "alice", "logs", "eliot.json"))
assert exists(join(temp_dir, "bob", "logs", "eliot.json"))
def test_alice_writes_bob_receives(magic_folder):
alice_dir, bob_dir = magic_folder
with open(join(alice_dir, "first_file"), "w") as f:
f.write("alice wrote this")
util.await_file_contents(join(bob_dir, "first_file"), "alice wrote this")
return
def test_alice_writes_bob_receives_multiple(magic_folder):
"""
When Alice does a series of updates, Bob should just receive them
with no .backup or .conflict files being produced.
"""
alice_dir, bob_dir = magic_folder
unwanted_files = [
join(bob_dir, "multiple.backup"),
join(bob_dir, "multiple.conflict")
]
# first update
with open(join(alice_dir, "multiple"), "w") as f:
f.write("alice wrote this")
util.await_file_contents(
join(bob_dir, "multiple"), "alice wrote this",
error_if=unwanted_files,
)
# second update
with open(join(alice_dir, "multiple"), "w") as f:
f.write("someone changed their mind")
util.await_file_contents(
join(bob_dir, "multiple"), "someone changed their mind",
error_if=unwanted_files,
)
# third update
with open(join(alice_dir, "multiple"), "w") as f:
f.write("absolutely final version ship it")
util.await_file_contents(
join(bob_dir, "multiple"), "absolutely final version ship it",
error_if=unwanted_files,
)
# forth update, but both "at once" so one should conflict
time.sleep(2)
with open(join(alice_dir, "multiple"), "w") as f:
f.write("okay one more attempt")
with open(join(bob_dir, "multiple"), "w") as f:
f.write("...but just let me add")
bob_conflict = join(bob_dir, "multiple.conflict")
alice_conflict = join(alice_dir, "multiple.conflict")
found = util.await_files_exist([
bob_conflict,
alice_conflict,
])
assert len(found) > 0, "Should have found a conflict"
print("conflict found (as expected)")
def test_alice_writes_bob_receives_old_timestamp(magic_folder):
alice_dir, bob_dir = magic_folder
fname = join(alice_dir, "ts_file")
ts = time.time() - (60 * 60 * 36) # 36 hours ago
with open(fname, "w") as f:
f.write("alice wrote this")
utime(fname, (time.time(), ts))
fname = join(bob_dir, "ts_file")
util.await_file_contents(fname, "alice wrote this")
# make sure the timestamp is correct
assert int(getmtime(fname)) == int(ts)
return
def test_bob_writes_alice_receives(magic_folder):
alice_dir, bob_dir = magic_folder
with open(join(bob_dir, "second_file"), "w") as f:
f.write("bob wrote this")
util.await_file_contents(join(alice_dir, "second_file"), "bob wrote this")
return
def test_alice_deletes(magic_folder):
# alice writes a file, waits for bob to get it and then deletes it.
alice_dir, bob_dir = magic_folder
with open(join(alice_dir, "delfile"), "w") as f:
f.write("alice wrote this")
util.await_file_contents(join(bob_dir, "delfile"), "alice wrote this")
# bob has the file; now alices deletes it
unlink(join(alice_dir, "delfile"))
# bob should remove his copy, but preserve a backup
util.await_file_vanishes(join(bob_dir, "delfile"))
util.await_file_contents(join(bob_dir, "delfile.backup"), "alice wrote this")
return
def test_alice_creates_bob_edits(magic_folder):
alice_dir, bob_dir = magic_folder
# alice writes a file
with open(join(alice_dir, "editfile"), "w") as f:
f.write("alice wrote this")
util.await_file_contents(join(bob_dir, "editfile"), "alice wrote this")
# now bob edits it
with open(join(bob_dir, "editfile"), "w") as f:
f.write("bob says foo")
util.await_file_contents(join(alice_dir, "editfile"), "bob says foo")
def test_bob_creates_sub_directory(magic_folder):
alice_dir, bob_dir = magic_folder
# bob makes a sub-dir, with a file in it
mkdir(join(bob_dir, "subdir"))
with open(join(bob_dir, "subdir", "a_file"), "w") as f:
f.write("bob wuz here")
# alice gets it
util.await_file_contents(join(alice_dir, "subdir", "a_file"), "bob wuz here")
# now bob deletes it again
shutil.rmtree(join(bob_dir, "subdir"))
# alice should delete it as well
util.await_file_vanishes(join(alice_dir, "subdir", "a_file"))
# i *think* it's by design that the subdir won't disappear,
# because a "a_file.backup" should appear...
util.await_file_contents(join(alice_dir, "subdir", "a_file.backup"), "bob wuz here")
def test_bob_creates_alice_deletes_bob_restores(magic_folder):
alice_dir, bob_dir = magic_folder
# bob creates a file
with open(join(bob_dir, "boom"), "w") as f:
f.write("bob wrote this")
util.await_file_contents(
join(alice_dir, "boom"),
"bob wrote this"
)
# alice deletes it (so bob should as well .. but keep a backup)
unlink(join(alice_dir, "boom"))
util.await_file_vanishes(join(bob_dir, "boom"))
assert exists(join(bob_dir, "boom.backup"))
# bob restore it, with new contents
unlink(join(bob_dir, "boom.backup"))
with open(join(bob_dir, "boom"), "w") as f:
f.write("bob wrote this again, because reasons")
# XXX double-check this behavior is correct!
# alice sees bob's update, but marks it as a conflict (because
# .. she previously deleted it? does that really make sense)
util.await_file_contents(
join(alice_dir, "boom"),
"bob wrote this again, because reasons",
)
def test_bob_creates_alice_deletes_alice_restores(magic_folder):
alice_dir, bob_dir = magic_folder
# bob creates a file
with open(join(bob_dir, "boom2"), "w") as f:
f.write("bob wrote this")
util.await_file_contents(
join(alice_dir, "boom2"),
"bob wrote this"
)
# alice deletes it (so bob should as well)
unlink(join(alice_dir, "boom2"))
util.await_file_vanishes(join(bob_dir, "boom2"))
# alice restore it, with new contents
with open(join(alice_dir, "boom2"), "w") as f:
f.write("alice re-wrote this again, because reasons")
util.await_file_contents(
join(bob_dir, "boom2"),
"alice re-wrote this again, because reasons"
)
def test_bob_conflicts_with_alice_fresh(magic_folder):
# both alice and bob make a file at "the same time".
alice_dir, bob_dir = magic_folder
# either alice or bob will "win" by uploading to the DMD first.
with open(join(bob_dir, 'alpha'), 'w') as f0, open(join(alice_dir, 'alpha'), 'w') as f1:
f0.write("this is bob's alpha\n")
f1.write("this is alice's alpha\n")
# there should be conflicts
_bob_conflicts_alice_await_conflicts('alpha', alice_dir, bob_dir)
def test_bob_conflicts_with_alice_preexisting(magic_folder):
# both alice and bob edit a file at "the same time" (similar to
# above, but the file already exists before the edits)
alice_dir, bob_dir = magic_folder
# have bob create the file
with open(join(bob_dir, 'beta'), 'w') as f:
f.write("original beta (from bob)\n")
util.await_file_contents(join(alice_dir, 'beta'), "original beta (from bob)\n")
# both alice and bob now have a "beta" file, at version 0
# either alice or bob will "win" by uploading to the DMD first
# (however, they should both detect a conflict)
with open(join(bob_dir, 'beta'), 'w') as f:
f.write("this is bob's beta\n")
with open(join(alice_dir, 'beta'), 'w') as f:
f.write("this is alice's beta\n")
# both alice and bob should see a conflict
_bob_conflicts_alice_await_conflicts("beta", alice_dir, bob_dir)
def _bob_conflicts_alice_await_conflicts(name, alice_dir, bob_dir):
"""
shared code between _fresh and _preexisting conflict test
"""
found = util.await_files_exist(
[
join(bob_dir, '{}.conflict'.format(name)),
join(alice_dir, '{}.conflict'.format(name)),
],
)
assert len(found) >= 1, "should be at least one conflict"
assert open(join(bob_dir, name), 'r').read() == "this is bob's {}\n".format(name)
assert open(join(alice_dir, name), 'r').read() == "this is alice's {}\n".format(name)
alice_conflict = join(alice_dir, '{}.conflict'.format(name))
bob_conflict = join(bob_dir, '{}.conflict'.format(name))
if exists(bob_conflict):
assert open(bob_conflict, 'r').read() == "this is alice's {}\n".format(name)
if exists(alice_conflict):
assert open(alice_conflict, 'r').read() == "this is bob's {}\n".format(name)
@pytest_twisted.inlineCallbacks
def test_edmond_uploads_then_restarts(reactor, request, temp_dir, introducer_furl, flog_gatherer, storage_nodes):
"""
ticket 2880: if a magic-folder client uploads something, then
re-starts a spurious .backup file should not appear
"""
edmond_dir = join(temp_dir, 'edmond')
edmond = yield util._create_node(
reactor, request, temp_dir, introducer_furl, flog_gatherer,
"edmond", web_port="tcp:9985:interface=localhost",
storage=False,
)
magic_folder = join(temp_dir, 'magic-edmond')
mkdir(magic_folder)
created = False
# create a magic-folder
# (how can we know that the grid is ready?)
for _ in range(10): # try 10 times
try:
proto = util._CollectOutputProtocol()
transport = reactor.spawnProcess(
proto,
sys.executable,
[
sys.executable, '-m', 'allmydata.scripts.runner',
'magic-folder', 'create',
'--poll-interval', '2',
'--basedir', edmond_dir,
'magik:',
'edmond_magic',
magic_folder,
]
)
yield proto.done
created = True
break
except Exception as e:
print("failed to create magic-folder: {}".format(e))
time.sleep(1)
assert created, "Didn't create a magic-folder"
# to actually-start the magic-folder we have to re-start
edmond.signalProcess('TERM')
yield edmond._protocol.exited
time.sleep(1)
edmond = yield util._run_node(reactor, edmond._node_dir, request, 'Completed initial Magic Folder scan successfully')
# add a thing to the magic-folder
with open(join(magic_folder, "its_a_file"), "w") as f:
f.write("edmond wrote this")
# fixme, do status-update attempts in a loop below
time.sleep(5)
# let it upload; poll the HTTP magic-folder status API until it is
# uploaded
from allmydata.scripts.magic_folder_cli import _get_json_for_fragment
with open(join(edmond_dir, u'private', u'api_auth_token'), 'rb') as f:
token = f.read()
uploaded = False
for _ in range(10):
options = {
"node-url": open(join(edmond_dir, u'node.url'), 'r').read().strip(),
}
try:
magic_data = _get_json_for_fragment(
options,
'magic_folder?t=json',
method='POST',
post_args=dict(
t='json',
name='default',
token=token,
)
)
for mf in magic_data:
if mf['status'] == u'success' and mf['path'] == u'its_a_file':
uploaded = True
break
except Exception as e:
time.sleep(1)
assert uploaded, "expected to upload 'its_a_file'"
# re-starting edmond right now would "normally" trigger the 2880 bug
# kill edmond
edmond.signalProcess('TERM')
yield edmond._protocol.exited
time.sleep(1)
edmond = yield util._run_node(reactor, edmond._node_dir, request, 'Completed initial Magic Folder scan successfully')
# XXX how can we say for sure if we've waited long enough? look at
# tail of logs for magic-folder ... somethingsomething?
print("waiting 20 seconds to see if a .backup appears")
for _ in range(20):
assert exists(join(magic_folder, "its_a_file"))
assert not exists(join(magic_folder, "its_a_file.backup"))
time.sleep(1)
@pytest_twisted.inlineCallbacks
def test_alice_adds_files_while_bob_is_offline(reactor, request, temp_dir, magic_folder):
"""
Alice can add new files to a magic folder while Bob is offline. When Bob
comes back online his copy is updated to reflect the new files.
"""
alice_magic_dir, bob_magic_dir = magic_folder
alice_node_dir = join(temp_dir, "alice")
bob_node_dir = join(temp_dir, "bob")
# Take Bob offline.
yield util.cli(reactor, bob_node_dir, "stop")
# Create a couple files in Alice's local directory.
some_files = list(
(name * 3) + ".added-while-offline"
for name
in "xyz"
)
for name in some_files:
with open(join(alice_magic_dir, name), "w") as f:
f.write(name + " some content")
good = False
for i in range(15):
status = yield util.magic_folder_cli(reactor, alice_node_dir, "status")
good = status.count(".added-while-offline (36 B): good, version=0") == len(some_files) * 2
if good:
# We saw each file as having a local good state and a remote good
# state. That means we're ready to involve Bob.
break
else:
time.sleep(1.0)
assert good, (
"Timed out waiting for good Alice state. Last status:\n{}".format(status)
)
# Start Bob up again
magic_text = 'Completed initial Magic Folder scan successfully'
yield util._run_node(reactor, bob_node_dir, request, magic_text)
yield util.await_files_exist(
list(
join(bob_magic_dir, name)
for name
in some_files
),
await_all=True,
)
# Let it settle. It would be nicer to have a readable status output we
# could query. Parsing the current text format is more than I want to
# deal with right now.
time.sleep(1.0)
conflict_files = list(name + ".conflict" for name in some_files)
assert all(
list(
not exists(join(bob_magic_dir, name))
for name
in conflict_files
),
)

7
integration/test_servers_of_happiness.py
View File

@ -12,7 +12,7 @@ import pytest_twisted
@pytest_twisted.inlineCallbacks
def test_upload_immutable(reactor, temp_dir, introducer_furl, flog_gatherer, storage_nodes, request):
yield util._create_node(
edna = yield util._create_node(
reactor, request, temp_dir, introducer_furl, flog_gatherer, "edna",
web_port="tcp:9983:interface=localhost",
storage=False,
@ -20,13 +20,10 @@ def test_upload_immutable(reactor, temp_dir, introducer_furl, flog_gatherer, sto
happy=7,
total=10,
)
util.await_client_ready(edna)
node_dir = join(temp_dir, 'edna')
print("waiting 10 seconds unil we're maybe ready")
yield task.deferLater(reactor, 10, lambda: None)
# upload a file, which should fail because we have don't have 7
# storage servers (but happiness is set to 7)
proto = util._CollectOutputProtocol()

2
integration/test_tor.py
View File

@ -14,7 +14,7 @@ import pytest_twisted
import util
# see "conftest.py" for the fixtures (e.g. "magic_folder")
# see "conftest.py" for the fixtures (e.g. "tor_network")
@pytest_twisted.inlineCallbacks
def test_onion_service_storage(reactor, request, temp_dir, flog_gatherer, tor_network, tor_introducer_furl):

521
integration/test_web.py Normal file
View File

@ -0,0 +1,521 @@
"""
These tests were originally written to achieve some level of
coverage for the WebAPI functionality during Python3 porting (there
aren't many tests of the Web API period).
Most of the tests have cursory asserts and encode 'what the WebAPI did
at the time of testing' -- not necessarily a cohesive idea of what the
WebAPI *should* do in every situation. It's not clear the latter
exists anywhere, however.
"""
import sys
import time
import shutil
import json
import urllib2
from os import mkdir, unlink, utime
from os.path import join, exists, getmtime
import allmydata.uri
import util
import requests
import pytest_twisted
import html5lib
from bs4 import BeautifulSoup
def test_index(alice):
"""
we can download the index file
"""
util.web_get(alice, u"")
def test_index_json(alice):
"""
we can download the index file as json
"""
data = util.web_get(alice, u"", params={u"t": u"json"})
# it should be valid json
json.loads(data)
def test_upload_download(alice):
"""
upload a file, then download it via readcap
"""
FILE_CONTENTS = u"some contents"
readcap = util.web_post(
alice, u"uri",
data={
u"t": u"upload",
u"format": u"mdmf",
},
files={
u"file": FILE_CONTENTS,
},
)
readcap = readcap.strip()
data = util.web_get(
alice, u"uri",
params={
u"uri": readcap,
u"filename": u"boom",
}
)
assert data == FILE_CONTENTS
def test_put(alice):
"""
use PUT to create a file
"""
FILE_CONTENTS = b"added via PUT" * 20
resp = requests.put(
util.node_url(alice.node_dir, u"uri"),
data=FILE_CONTENTS,
)
cap = allmydata.uri.from_string(resp.text.strip().encode('ascii'))
cfg = alice.get_config()
assert isinstance(cap, allmydata.uri.CHKFileURI)
assert cap.size == len(FILE_CONTENTS)
assert cap.total_shares == int(cfg.get_config("client", "shares.total"))
assert cap.needed_shares == int(cfg.get_config("client", "shares.needed"))
def test_helper_status(storage_nodes):
"""
successfully GET the /helper_status page
"""
url = util.node_url(storage_nodes[0].node_dir, "helper_status")
resp = requests.get(url)
assert resp.status_code >= 200 and resp.status_code < 300
dom = BeautifulSoup(resp.content, "html5lib")
assert unicode(dom.h1.string) == u"Helper Status"
def test_deep_stats(alice):
"""
create a directory, do deep-stats on it and prove the /operations/
URIs work
"""
resp = requests.post(
util.node_url(alice.node_dir, "uri"),
params={
"format": "sdmf",
"t": "mkdir",
"redirect_to_result": "true",
},
)
assert resp.status_code >= 200 and resp.status_code < 300
# when creating a directory, we'll be re-directed to a URL
# containing our writecap..
uri = urllib2.unquote(resp.url)
assert 'URI:DIR2:' in uri
dircap = uri[uri.find("URI:DIR2:"):].rstrip('/')
dircap_uri = util.node_url(alice.node_dir, "uri/{}".format(urllib2.quote(dircap)))
# POST a file into this directory
FILE_CONTENTS = u"a file in a directory"
resp = requests.post(
dircap_uri,
data={
u"t": u"upload",
u"when_done": u".",
},
files={
u"file": FILE_CONTENTS,
},
)
# confirm the file is in the directory
resp = requests.get(
dircap_uri,
params={
u"t": u"json",
},
)
d = json.loads(resp.content)
k, data = d
assert k == u"dirnode"
assert len(data['children']) == 1
k, child = data['children'].values()[0]
assert k == u"filenode"
assert child['size'] == len(FILE_CONTENTS)
# perform deep-stats on it...
resp = requests.post(
dircap_uri,
data={
u"t": u"start-deep-stats",
u"ophandle": u"something_random",
},
)
assert resp.status_code >= 200 and resp.status_code < 300
# confirm we get information from the op .. after its done
tries = 10
while tries > 0:
tries -= 1
resp = requests.get(
util.node_url(alice.node_dir, u"operations/something_random"),
)
d = json.loads(resp.content)
if d['size-literal-files'] == len(FILE_CONTENTS):
print("stats completed successfully")
break
else:
print("{} != {}; waiting".format(d['size-literal-files'], len(FILE_CONTENTS)))
time.sleep(.5)
def test_status(alice):
"""
confirm we get something sensible from /status and the various sub-types
"""
# upload a file
# (because of the nature of the integration-tests, we can only
# assert things about "our" file because we don't know what other
# operations may have happened in the grid before our test runs).
FILE_CONTENTS = u"all the Important Data of alice\n" * 1200
resp = requests.put(
util.node_url(alice.node_dir, u"uri"),
data=FILE_CONTENTS,
)
cap = resp.text.strip()
print("Uploaded data, cap={}".format(cap))
resp = requests.get(
util.node_url(alice.node_dir, u"uri/{}".format(urllib2.quote(cap))),
)
print("Downloaded {} bytes of data".format(len(resp.content)))
assert resp.content == FILE_CONTENTS
resp = requests.get(
util.node_url(alice.node_dir, "status"),
)
dom = html5lib.parse(resp.content)
hrefs = [
a.get('href')
for a in dom.iter(u'{http://www.w3.org/1999/xhtml}a')
]
found_upload = False
found_download = False
for href in hrefs:
if href.startswith(u"/") or not href:
continue
resp = requests.get(
util.node_url(alice.node_dir, u"status/{}".format(href)),
)
if href.startswith(u'up'):
assert "File Upload Status" in resp.content
if "Total Size: {}".format(len(FILE_CONTENTS)) in resp.content:
found_upload = True
elif href.startswith(u'down'):
assert "File Download Status" in resp.content
if "Total Size: {}".format(len(FILE_CONTENTS)) in resp.content:
found_download = True
# download the specialized event information
resp = requests.get(
util.node_url(alice.node_dir, u"status/{}/event_json".format(href)),
)
js = json.loads(resp.content)
# there's usually just one "read" operation, but this can handle many ..
total_bytes = sum([st['bytes_returned'] for st in js['read']], 0)
assert total_bytes == len(FILE_CONTENTS)
assert found_upload, "Failed to find the file we uploaded in the status-page"
assert found_download, "Failed to find the file we downloaded in the status-page"
def test_directory_deep_check(alice):
"""
use deep-check and confirm the result pages work
"""
# create a directory
resp = requests.post(
util.node_url(alice.node_dir, u"uri"),
params={
u"t": u"mkdir",
u"redirect_to_result": u"true",
}
)
# get json information about our directory
dircap_url = resp.url
resp = requests.get(
dircap_url,
params={u"t": u"json"},
)
dir_meta = json.loads(resp.content)
# upload a file of pangrams into the directory
FILE_CONTENTS = u"Sphinx of black quartz, judge my vow.\n" * (2048*10)
resp = requests.post(
dircap_url,
params={
u"t": u"upload",
u"upload-chk": u"upload-chk",
},
files={
u"file": FILE_CONTENTS,
}
)
cap0 = resp.content
print("Uploaded data0, cap={}".format(cap0))
# a different pangram
FILE_CONTENTS = u"The five boxing wizards jump quickly.\n" * (2048*10)
resp = requests.post(
dircap_url,
params={
u"t": u"upload",
u"upload-chk": u"upload-chk",
},
files={
u"file": FILE_CONTENTS,
}
)
cap1 = resp.content
print("Uploaded data1, cap={}".format(cap1))
resp = requests.get(
util.node_url(alice.node_dir, u"uri/{}".format(urllib2.quote(cap0))),
params={u"t": u"info"},
)
def check_repair_data(checkdata):
assert checkdata["healthy"] is True
assert checkdata["count-happiness"] == 4
assert checkdata["count-good-share-hosts"] == 4
assert checkdata["count-shares-good"] == 4
assert checkdata["count-corrupt-shares"] == 0
assert checkdata["list-corrupt-shares"] == []
# do a "check" (once for HTML, then with JSON for easier asserts)
resp = requests.post(
dircap_url,
params={
u"t": u"check",
u"return_to": u".",
u"verify": u"true",
}
)
resp = requests.post(
dircap_url,
params={
u"t": u"check",
u"return_to": u".",
u"verify": u"true",
u"output": u"JSON",
}
)
check_repair_data(json.loads(resp.content)["results"])
# "check and repair"
resp = requests.post(
dircap_url,
params={
u"t": u"check",
u"return_to": u".",
u"verify": u"true",
u"repair": u"true",
}
)
resp = requests.post(
dircap_url,
params={
u"t": u"check",
u"return_to": u".",
u"verify": u"true",
u"repair": u"true",
u"output": u"JSON",
}
)
check_repair_data(json.loads(resp.content)["post-repair-results"]["results"])
# start a "deep check and repair"
resp = requests.post(
dircap_url,
params={
u"t": u"start-deep-check",
u"return_to": u".",
u"verify": u"on",
u"repair": u"on",
u"output": u"JSON",
u"ophandle": u"deadbeef",
}
)
deepcheck_uri = resp.url
data = json.loads(resp.content)
tries = 10
while not data['finished'] and tries > 0:
tries -= 1
time.sleep(0.5)
print("deep-check not finished, reloading")
resp = requests.get(deepcheck_uri, params={u"output": "JSON"})
data = json.loads(resp.content)
print("deep-check finished")
assert data[u"stats"][u"count-immutable-files"] == 1
assert data[u"stats"][u"count-literal-files"] == 0
assert data[u"stats"][u"largest-immutable-file"] == 778240
assert data[u"count-objects-checked"] == 2
# also get the HTML version
resp = requests.post(
dircap_url,
params={
u"t": u"start-deep-check",
u"return_to": u".",
u"verify": u"on",
u"repair": u"on",
u"ophandle": u"definitely_random",
}
)
deepcheck_uri = resp.url
# if the operations isn't done, there's an <H2> tag with the
# reload link; otherwise there's only an <H1> tag..wait up to 5
# seconds for this to respond properly.
for _ in range(5):
resp = requests.get(deepcheck_uri)
dom = BeautifulSoup(resp.content, "html5lib")
if dom.h1 and u'Results' in unicode(dom.h1.string):
break
if dom.h2 and dom.h2.a and u"Reload" in unicode(dom.h2.a.string):
dom = None
time.sleep(1)
assert dom is not None, "Operation never completed"
def test_storage_info(storage_nodes):
"""
retrieve and confirm /storage URI for one storage node
"""
storage0 = storage_nodes[0]
requests.get(
util.node_url(storage0.node_dir, u"storage"),
)
def test_storage_info_json(storage_nodes):
"""
retrieve and confirm /storage?t=json URI for one storage node
"""
storage0 = storage_nodes[0]
resp = requests.get(
util.node_url(storage0.node_dir, u"storage"),
params={u"t": u"json"},
)
data = json.loads(resp.content)
assert data[u"stats"][u"storage_server.reserved_space"] == 1000000000
def test_introducer_info(introducer):
"""
retrieve and confirm /introducer URI for the introducer
"""
resp = requests.get(
util.node_url(introducer.node_dir, u""),
)
assert "Introducer" in resp.content
resp = requests.get(
util.node_url(introducer.node_dir, u""),
params={u"t": u"json"},
)
data = json.loads(resp.content)
assert "announcement_summary" in data
assert "subscription_summary" in data
def test_mkdir_with_children(alice):
"""
create a directory using ?t=mkdir-with-children
"""
# create a file to put in our directory
FILE_CONTENTS = u"some file contents\n" * 500
resp = requests.put(
util.node_url(alice.node_dir, u"uri"),
data=FILE_CONTENTS,
)
filecap = resp.content.strip()
# create a (sub) directory to put in our directory
resp = requests.post(
util.node_url(alice.node_dir, u"uri"),
params={
u"t": u"mkdir",
}
)
# (we need both the read-write and read-only URIs I guess)
dircap = resp.content
dircap_obj = allmydata.uri.from_string(dircap)
dircap_ro = dircap_obj.get_readonly().to_string()
# create json information about our directory
meta = {
"a_file": [
"filenode", {
"ro_uri": filecap,
"metadata": {
"ctime": 1202777696.7564139,
"mtime": 1202777696.7564139,
"tahoe": {
"linkcrtime": 1202777696.7564139,
"linkmotime": 1202777696.7564139
}
}
}
],
"some_subdir": [
"dirnode", {
"rw_uri": dircap,
"ro_uri": dircap_ro,
"metadata": {
"ctime": 1202778102.7589991,
"mtime": 1202778111.2160511,
"tahoe": {
"linkcrtime": 1202777696.7564139,
"linkmotime": 1202777696.7564139
}
}
}
]
}
# create a new directory with one file and one sub-dir (all-at-once)
resp = util.web_post(
alice, u"uri",
params={u"t": "mkdir-with-children"},
data=json.dumps(meta),
)
assert resp.startswith("URI:DIR2")
cap = allmydata.uri.from_string(resp)
assert isinstance(cap, allmydata.uri.DirectoryURI)

229
integration/util.py
View File

@ -1,5 +1,6 @@
import sys
import time
import json
from os import mkdir
from os.path import exists, join
from six.moves import StringIO
@ -10,11 +11,14 @@ from twisted.internet.defer import Deferred, succeed
from twisted.internet.protocol import ProcessProtocol
from twisted.internet.error import ProcessExitedAlready, ProcessDone
import requests
from allmydata.util.configutil import (
get_config,
set_config,
write_config,
)
from allmydata import client
import pytest_twisted
@ -127,19 +131,19 @@ class _MagicTextProtocol(ProcessProtocol):
sys.stdout.write(data)
def _cleanup_twistd_process(twistd_process, exited):
def _cleanup_tahoe_process(tahoe_transport, exited):
"""
Terminate the given process with a kill signal (SIGKILL on POSIX,
TerminateProcess on Windows).
:param twistd_process: The `IProcessTransport` representing the process.
:param tahoe_transport: The `IProcessTransport` representing the process.
:param exited: A `Deferred` which fires when the process has exited.
:return: After the process has exited.
"""
try:
print("signaling {} with KILL".format(twistd_process.pid))
twistd_process.signalProcess('KILL')
print("signaling {} with TERM".format(tahoe_transport.pid))
tahoe_transport.signalProcess('TERM')
print("signaled, blocking on exit")
pytest_twisted.blockon(exited)
print("exited, goodbye")
@ -147,20 +151,68 @@ def _cleanup_twistd_process(twistd_process, exited):
pass
def run_tahoe(reactor, *args, **kwargs):
def run_tahoe(reactor, request, *args):
"""
Helper to run tahoe with optional coverage
"""
stdin = kwargs.get('stdin', None)
protocol = _CollectOutputProtocol(stdin=stdin)
process = reactor.spawnProcess(
protocol,
sys.executable,
(sys.executable, '-m', 'allmydata.scripts.runner') + args
)
process = _tahoe_runner_optional_coverage(protocol, reactor, request, args)
process.exited = protocol.done
return protocol.done
def _tahoe_runner_optional_coverage(proto, reactor, request, other_args):
"""
Internal helper. Calls spawnProcess with `-m
allmydata.scripts.runner` and `other_args`, optionally inserting a
`--coverage` option if the `request` indicates we should.
"""
if request.config.getoption('coverage'):
args = [sys.executable, '-m', 'coverage', 'run', '-m', 'allmydata.scripts.runner', '--coverage']
else:
args = [sys.executable, '-m', 'allmydata.scripts.runner']
args += other_args
return reactor.spawnProcess(
proto,
sys.executable,
args,
)
class TahoeProcess(object):
"""
A running Tahoe process, with associated information.
"""
def __init__(self, process_transport, node_dir):
self._process_transport = process_transport # IProcessTransport instance
self._node_dir = node_dir # path
@property
def transport(self):
return self._process_transport
@property
def node_dir(self):
return self._node_dir
def get_config(self):
return client.read_config(
self._node_dir,
u"portnum",
)
def __str__(self):
return "<TahoeProcess in '{}'>".format(self._node_dir)
def _run_node(reactor, node_dir, request, magic_text):
"""
Run a tahoe process from its node_dir.
:returns: a TahoeProcess for this node
"""
if magic_text is None:
magic_text = "client running"
protocol = _MagicTextProtocol(magic_text)
@ -168,27 +220,29 @@ def _run_node(reactor, node_dir, request, magic_text):
# on windows, "tahoe start" means: run forever in the foreground,
# but on linux it means daemonize. "tahoe run" is consistent
# between platforms.
process = reactor.spawnProcess(
transport = _tahoe_runner_optional_coverage(
protocol,
sys.executable,
(
sys.executable, '-m', 'allmydata.scripts.runner',
reactor,
request,
[
'--eliot-destination', 'file:{}/logs/eliot.json'.format(node_dir),
'run',
node_dir,
),
],
)
process.exited = protocol.exited
transport.exited = protocol.exited
request.addfinalizer(partial(_cleanup_twistd_process, process, protocol.exited))
request.addfinalizer(partial(_cleanup_tahoe_process, transport, protocol.exited))
# we return the 'process' ITransport instance
# XXX abusing the Deferred; should use .when_magic_seen() or something?
# XXX abusing the Deferred; should use .when_magic_seen() pattern
def got_proto(proto):
process._protocol = proto
process._node_dir = node_dir
return process
transport._protocol = proto
return TahoeProcess(
transport,
node_dir,
)
protocol.magic_seen.addCallback(got_proto)
return protocol.magic_seen
@ -213,7 +267,6 @@ def _create_node(reactor, request, temp_dir, introducer_furl, flog_gatherer, nam
mkdir(node_dir)
done_proto = _ProcessExitedProtocol()
args = [
sys.executable, '-m', 'allmydata.scripts.runner',
'create-node',
'--nickname', name,
'--introducer', introducer_furl,
@ -223,16 +276,13 @@ def _create_node(reactor, request, temp_dir, introducer_furl, flog_gatherer, nam
'--shares-needed', unicode(needed),
'--shares-happy', unicode(happy),
'--shares-total', unicode(total),
'--helper',
]
if not storage:
args.append('--no-storage')
args.append(node_dir)
reactor.spawnProcess(
done_proto,
sys.executable,
args,
)
_tahoe_runner_optional_coverage(done_proto, reactor, request, args)
created_d = done_proto.done
def created(_):
@ -365,17 +415,118 @@ def await_file_vanishes(path, timeout=10):
raise FileShouldVanishException(path, timeout)
def cli(reactor, node_dir, *argv):
def cli(request, reactor, node_dir, *argv):
"""
Run a tahoe CLI subcommand for a given node, optionally running
under coverage if '--coverage' was supplied.
"""
proto = _CollectOutputProtocol()
reactor.spawnProcess(
proto,
sys.executable,
[
sys.executable, '-m', 'allmydata.scripts.runner',
'--node-directory', node_dir,
] + list(argv),
_tahoe_runner_optional_coverage(
proto, reactor, request,
['--node-directory', node_dir] + list(argv),
)
return proto.done
def magic_folder_cli(reactor, node_dir, *argv):
return cli(reactor, node_dir, "magic-folder", *argv)
def node_url(node_dir, uri_fragment):
"""
Create a fully qualified URL by reading config from `node_dir` and
adding the `uri_fragment`
"""
with open(join(node_dir, "node.url"), "r") as f:
base = f.read().strip()
url = base + uri_fragment
return url
def _check_status(response):
"""
Check the response code is a 2xx (raise an exception otherwise)
"""
if response.status_code < 200 or response.status_code >= 300:
raise ValueError(
"Expected a 2xx code, got {}".format(response.status_code)
)
def web_get(tahoe, uri_fragment, **kwargs):
"""
Make a GET request to the webport of `tahoe` (a `TahoeProcess`,
usually from a fixture (e.g. `alice`). This will look like:
`http://localhost:<webport>/<uri_fragment>`. All `kwargs` are
passed on to `requests.get`
"""
url = node_url(tahoe.node_dir, uri_fragment)
resp = requests.get(url, **kwargs)
_check_status(resp)
return resp.content
def web_post(tahoe, uri_fragment, **kwargs):
"""
Make a POST request to the webport of `node` (a `TahoeProcess,
usually from a fixture e.g. `alice`). This will look like:
`http://localhost:<webport>/<uri_fragment>`. All `kwargs` are
passed on to `requests.post`
"""
url = node_url(tahoe.node_dir, uri_fragment)
resp = requests.post(url, **kwargs)
_check_status(resp)
return resp.content
def await_client_ready(tahoe, timeout=10, liveness=60*2):
"""
Uses the status API to wait for a client-type node (in `tahoe`, a
`TahoeProcess` instance usually from a fixture e.g. `alice`) to be
'ready'. A client is deemed ready if:
- it answers `http://<node_url>/statistics/?t=json/`
- there is at least one storage-server connected
- every storage-server has a "last_received_data" and it is
within the last `liveness` seconds
We will try for up to `timeout` seconds for the above conditions
to be true. Otherwise, an exception is raised
"""
start = time.time()
while (time.time() - start) < float(timeout):
try:
data = web_get(tahoe, u"", params={u"t": u"json"})
js = json.loads(data)
except Exception as e:
print("waiting because '{}'".format(e))
time.sleep(1)
continue
if len(js['servers']) == 0:
print("waiting because no servers at all")
time.sleep(1)
continue
server_times = [
server['last_received_data']
for server in js['servers']
]
# if any times are null/None that server has never been
# contacted (so it's down still, probably)
if any(t is None for t in server_times):
print("waiting because at least one server not contacted")
time.sleep(1)
continue
# check that all times are 'recent enough'
if any([time.time() - t > liveness for t in server_times]):
print("waiting because at least one server too old")
time.sleep(1)
continue
# we have a status with at least one server, and all servers
# have been contacted recently
return True
# we only fall out of the loop when we've timed out
raise RuntimeError(
"Waited {} seconds for {} to be 'ready' but it never was".format(
timeout,
tahoe,
)
)

39
misc/build_helpers/run-deprecations.py
View File

@ -1,6 +1,6 @@
from __future__ import print_function
import sys, os, io
import sys, os, io, re
from twisted.internet import reactor, protocol, task, defer
from twisted.python.procutils import which
from twisted.python import usage
@ -12,6 +12,7 @@ from twisted.python import usage
class Options(usage.Options):
optParameters = [
["warnings", None, None, "file to write warnings into at end of test run"],
["package", None, None, "Python package to which to restrict warning collection"]
]
def parseArgs(self, command, *args):
@ -19,7 +20,7 @@ class Options(usage.Options):
self["args"] = list(args)
description = """Run as:
PYTHONWARNINGS=default::DeprecationWarning python run-deprecations.py [--warnings=STDERRFILE] COMMAND ARGS..
PYTHONWARNINGS=default::DeprecationWarning python run-deprecations.py [--warnings=STDERRFILE] [--package=PYTHONPACKAGE ] COMMAND ARGS..
"""
class RunPP(protocol.ProcessProtocol):
@ -34,6 +35,34 @@ class RunPP(protocol.ProcessProtocol):
rc = reason.value.exitCode
self.d.callback((signal, rc))
def make_matcher(options):
"""
Make a function that matches a line with a relevant deprecation.
A deprecation warning line looks something like this::
somepath/foo/bar/baz.py:43: DeprecationWarning: Foo is deprecated, try bar instead.
Sadly there is no guarantee warnings begin at the beginning of a line
since they are written to output without coordination with whatever other
Python code is running in the process.
:return: A one-argument callable that accepts a string and returns
``True`` if it contains an interesting warning and ``False``
otherwise.
"""
pattern = r".*\.py[oc]?:\d+:" # (Pending)?DeprecationWarning: .*"
if options["package"]:
pattern = r".*/{}/".format(
re.escape(options["package"]),
) + pattern
expression = re.compile(pattern)
def match(line):
return expression.match(line) is not None
return match
@defer.inlineCallbacks
def run_command(main):
config = Options()
@ -63,6 +92,8 @@ def run_command(main):
reactor.spawnProcess(pp, exe, [exe] + config["args"], env=None)
(signal, rc) = yield pp.d
match = make_matcher(config)
# maintain ordering, but ignore duplicates (for some reason, either the
# 'warnings' module or twisted.python.deprecate isn't quashing them)
already = set()
@ -75,12 +106,12 @@ def run_command(main):
pp.stdout.seek(0)
for line in pp.stdout.readlines():
if "DeprecationWarning" in line:
if match(line):
add(line) # includes newline
pp.stderr.seek(0)
for line in pp.stderr.readlines():
if "DeprecationWarning" in line:
if match(line):
add(line)
if warnings:

3
misc/build_helpers/show-tool-versions.py
View File

@ -146,8 +146,7 @@ print_py_pkg_ver('mock')
print_py_pkg_ver('Nevow', 'nevow')
print_py_pkg_ver('pyasn1')
print_py_pkg_ver('pycparser')
print_py_pkg_ver('pycrypto', 'Crypto')
print_py_pkg_ver('pycryptopp')
print_py_pkg_ver('cryptography')
print_py_pkg_ver('pyflakes')
print_py_pkg_ver('pyOpenSSL', 'OpenSSL')
print_py_pkg_ver('six')

2
misc/build_helpers/test-osx-pkg.py
View File

@ -15,7 +15,6 @@
# allmydata-tahoe: 1.10.0.post185.dev0 [2249-deps-and-osx-packaging-1: 76ac53846042d9a4095995be92af66cdc09d5ad0-dirty] (/Applications/tahoe.app/src)
# foolscap: 0.7.0 (/Applications/tahoe.app/support/lib/python2.7/site-packages/foolscap-0.7.0-py2.7.egg)
# pycryptopp: 0.6.0.1206569328141510525648634803928199668821045408958 (/Applications/tahoe.app/support/lib/python2.7/site-packages/pycryptopp-0.6.0.1206569328141510525648634803928199668821045408958-py2.7-macosx-10.9-intel.egg)
# zfec: 1.4.24 (/Applications/tahoe.app/support/lib/python2.7/site-packages/zfec-1.4.24-py2.7-macosx-10.9-intel.egg)
# Twisted: 13.0.0 (/Applications/tahoe.app/support/lib/python2.7/site-packages/Twisted-13.0.0-py2.7-macosx-10.9-intel.egg)
# Nevow: 0.11.1 (/Applications/tahoe.app/support/lib/python2.7/site-packages/Nevow-0.11.1-py2.7.egg)
@ -23,7 +22,6 @@
# python: 2.7.5 (/usr/bin/python)
# platform: Darwin-13.4.0-x86_64-i386-64bit (None)
# pyOpenSSL: 0.13 (/System/Library/Frameworks/Python.framework/Versions/2.7/Extras/lib/python)
# pycrypto: 2.6.1 (/Applications/tahoe.app/support/lib/python2.7/site-packages/pycrypto-2.6.1-py2.7-macosx-10.9-intel.egg)
# pyasn1: 0.1.7 (/Applications/tahoe.app/support/lib/python2.7/site-packages/pyasn1-0.1.7-py2.7.egg)
# mock: 1.0.1 (/Applications/tahoe.app/support/lib/python2.7/site-packages)
# setuptools: 0.6c16dev6 (/Applications/tahoe.app/support/lib/python2.7/site-packages/setuptools-0.6c16dev6.egg)

2
misc/coding_tools/check-miscaptures.py
View File

@ -138,7 +138,7 @@ def report(out, path, results):
print(path + (":%r %s captures %r assigned at line %d" % r), file=out)
def check(sources, out):
class Counts:
class Counts(object):
n = 0
processed_files = 0
suspect_files = 0

4
misc/operations_helpers/provisioning/reliability.py
View File

@ -8,7 +8,7 @@ DAY=24*60*60
MONTH=31*DAY
YEAR=365*DAY
class ReliabilityModel:
class ReliabilityModel(object):
"""Generate a model of system-wide reliability, given several input
parameters.
@ -207,7 +207,7 @@ class ReliabilityModel:
repair = matrix(new_repair_rows)
return repair
class ReliabilityReport:
class ReliabilityReport(object):
def __init__(self):
self.samples = []

2
misc/operations_helpers/provisioning/test_provisioning.py
View File

@ -10,7 +10,7 @@ except ImportError:
from nevow import inevow
from zope.interface import implements
class MyRequest:
class MyRequest(object):
implements(inevow.IRequest)
pass

34
misc/python3/depgraph.sh Executable file
View File

@ -0,0 +1,34 @@
#!/usr/bin/env bash
set -x
set -eo pipefail
TAHOE="${PWD}"
git clone -b gh-pages git@github.com:tahoe-lafs/tahoe-depgraph.git
cd tahoe-depgraph
# Generate the maybe-changed data.
python "${TAHOE}"/misc/python3/tahoe-depgraph.py "${TAHOE}"
if git diff-index --quiet HEAD; then
echo "Declining to commit without any changes."
exit 0
fi
git config user.name 'Build Automation'
git config user.email 'tahoe-dev@tahoe-lafs.org'
git add tahoe-deps.json tahoe-ported.json
git commit -m "\
Built from ${CIRCLE_REPOSITORY_URL}@${CIRCLE_SHA1}
tahoe-depgraph was $(git rev-parse HEAD)
"
if [ "${CIRCLE_BRANCH}" != "master" ]; then
echo "Declining to update dependency graph for non-master build."
exit 0
fi
# Publish it on GitHub.
git push -q origin gh-pages

123
misc/python3/tahoe-depgraph.py Normal file
View File

@ -0,0 +1,123 @@
# Copyright 2004, 2009 Toby Dickenson
# Copyright 2014-2015 Aaron Gallagher
#
# Permission is hereby granted, free of charge, to any person obtaining
# a copy of this software and associated documentation files (the
# "Software"), to deal in the Software without restriction, including
# without limitation the rights to use, copy, modify, merge, publish,
# distribute, sublicense, and/or sell copies of the Software, and to
# permit persons to whom the Software is furnished to do so, subject
# to the following conditions:
#
# The above copyright notice and this permission notice shall be included
# in all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
# IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
# CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
# TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
# SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
import collections
import functools
import json
import os
import modulefinder
import sys
import tempfile
from twisted.python import reflect
class mymf(modulefinder.ModuleFinder):
def __init__(self, *args, **kwargs):
self._depgraph = collections.defaultdict(set)
self._types = {}
self._last_caller = None
modulefinder.ModuleFinder.__init__(self, *args, **kwargs)
def import_hook(self, name, caller=None, fromlist=None, level=None):
old_last_caller = self._last_caller
try:
self._last_caller = caller
return modulefinder.ModuleFinder.import_hook(
self, name, caller, fromlist)
finally:
self._last_caller = old_last_caller
def import_module(self, partnam, fqname, parent):
if partnam.endswith('_py3'):
return None
r = modulefinder.ModuleFinder.import_module(
self, partnam, fqname, parent)
last_caller = self._last_caller
if r is not None and 'allmydata' in r.__name__:
if last_caller is None or last_caller.__name__ == '__main__':
self._depgraph[fqname]
else:
self._depgraph[last_caller.__name__].add(fqname)
return r
def load_module(self, fqname, fp, pathname, (suffix, mode, type)):
r = modulefinder.ModuleFinder.load_module(
self, fqname, fp, pathname, (suffix, mode, type))
if r is not None:
self._types[r.__name__] = type
return r
def as_json(self):
return {
'depgraph': {
name: dict.fromkeys(deps, 1)
for name, deps in self._depgraph.iteritems()},
'types': self._types,
}
json_dump = functools.partial(
json.dump, indent=4, separators=(',', ': '), sort_keys=True)
def main(target):
mf = mymf(sys.path[:], 0, [])
moduleNames = []
for path, dirnames, filenames in os.walk(os.path.join(target, 'src', 'allmydata')):
if 'test' in dirnames:
dirnames.remove('test')
for filename in filenames:
if not filename.endswith('.py'):
continue
if filename in ('setup.py',):
continue
if '-' in filename:
# a script like update-documentation.py
continue
if filename != '__init__.py':
filepath = os.path.join(path, filename)
else:
filepath = path
moduleNames.append(reflect.filenameToModuleName(filepath))
with tempfile.NamedTemporaryFile() as tmpfile:
for moduleName in moduleNames:
tmpfile.write('import %s\n' % moduleName)
tmpfile.flush()
mf.run_script(tmpfile.name)
with open('tahoe-deps.json', 'wb') as outfile:
json_dump(mf.as_json(), outfile)
outfile.write('\n')
ported_modules_path = os.path.join(target, "src", "allmydata", "ported-modules.txt")
with open(ported_modules_path) as ported_modules:
port_status = dict.fromkeys((line.strip() for line in ported_modules), "ported")
with open('tahoe-ported.json', 'wb') as outfile:
json_dump(port_status, outfile)
outfile.write('\n')
if __name__ == '__main__':
main(*sys.argv[1:])

4
misc/simulators/ringsim.py
View File

@ -54,7 +54,7 @@ print("average file size:", abbreviate_space(avg_filesize))
SERVER_CAPACITY = 10**12
class Server:
class Server(object):
def __init__(self, nodeid, capacity):
self.nodeid = nodeid
self.used = 0
@ -75,7 +75,7 @@ class Server:
else:
return "<%s %s>" % (self.__class__.__name__, self.nodeid)
class Ring:
class Ring(object):
SHOW_MINMAX = False
def __init__(self, numservers, seed, permute):
self.servers = []

2
misc/simulators/simulate_load.py
View File

@ -8,7 +8,7 @@ import random
SERVER_CAPACITY = 10**12
class Server:
class Server(object):
def __init__(self):
self.si = random.randrange(0, 2**31)
self.used = 0

6
misc/simulators/simulator.py
View File

@ -17,7 +17,7 @@ def sha(s):
def randomid():
return os.urandom(20)
class Node:
class Node(object):
def __init__(self, nid, introducer, simulator):
self.nid = nid
self.introducer = introducer
@ -112,7 +112,7 @@ class Node:
self.introducer.delete(fileid)
return True
class Introducer:
class Introducer(object):
def __init__(self, simulator):
self.living_files = {}
self.utilization = 0 # total size of all active files
@ -149,7 +149,7 @@ class Introducer:
self.simulator.stamp_utilization(self.utilization)
del self.living_files[fileid]
class Simulator:
class Simulator(object):
NUM_NODES = 1000
EVENTS = ["ADDFILE", "DELFILE", "ADDNODE", "DELNODE"]
RATE_ADDFILE = 1.0 / 10

2
misc/simulators/sizes.py
View File

@ -37,7 +37,7 @@ GiB=1024*MiB
TiB=1024*GiB
PiB=1024*TiB
class Sizes:
class Sizes(object):
def __init__(self, mode, file_size, arity=2):
MAX_SEGSIZE = 128*KiB
self.mode = mode

2
misc/simulators/storage-overhead.py
View File

@ -12,7 +12,7 @@ def roundup(size, blocksize=4096):
return blocksize * mathutil.div_ceil(size, blocksize)
class BigFakeString:
class BigFakeString(object):
def __init__(self, length):
self.length = length
self.fp = 0

1
newsfragments/1432.feature
View File

@ -1 +0,0 @@
Magic-Folders are now supported on macOS.

1
newsfragments/2870.bugfix
View File

@ -1 +0,0 @@
refactor initialization code to be more async-friendly

1
newsfragments/2908.other
View File

@ -1 +0,0 @@
Tahoe-LAFS now uses towncrier to maintain the NEWS file.

1
newsfragments/2920.other
View File

@ -1 +0,0 @@
The release process document has been updated.

1
newsfragments/2933.other
View File

@ -1 +0,0 @@
allmydata.test.test_system.SystemTest is now more reliable with respect to bound address collisions.

1
newsfragments/2935.bugfix
View File

@ -1 +0,0 @@
Configuration-checking code wasn't being called due to indenting

1
newsfragments/2936.bugfix
View File

@ -1 +0,0 @@
refactor configuration handling out of Node into _Config

1
newsfragments/2944.bugfix
View File

@ -1 +0,0 @@
Updated the Tor release key, used by the integration tests.

1
newsfragments/2950.bugfix
View File

@ -1 +0,0 @@
`tahoe backup` no longer fails with an unhandled exception when it encounters a special file (device, fifo) in the backup source.

1
newsfragments/2955.feature
View File

@ -1 +0,0 @@
Fedora 29 is now tested as part of the project's continuous integration system.

1
newsfragments/2955.removed
View File

@ -1 +0,0 @@
Fedora 27 is no longer tested as part of the project's continuous integration system.

1
newsfragments/2956.other
View File

@ -1 +0,0 @@
The Tox configuration has been fixed to work around a problem on Windows CI.

1
newsfragments/2957.installation
View File

@ -1 +0,0 @@
Tahoe-LAFS now depends on Twisted 16.6 or newer.

1
newsfragments/2958.other
View File

@ -1 +0,0 @@
The PyInstaller CI job now works around a pip/pyinstaller incompatibility.

1
newsfragments/2959.other
View File

@ -1 +0,0 @@
Some CI jobs for integration tests have been moved from TravisCI to CircleCI.

1
newsfragments/2960.other
View File

@ -1 +0,0 @@
Several warnings from a new release of pyflakes have been fixed.

1
newsfragments/2961.other
View File

@ -1 +0,0 @@
Some Slackware 14.2 continuous integration problems have been resolved.

1
newsfragments/2962.other
View File

@ -1 +0,0 @@
Some macOS continuous integration failures have been fixed.

1
newsfragments/2965.bugfix
View File

@ -1 +0,0 @@
Magic-Folders now creates spurious conflict files in fewer cases. In particular, if files are added to the folder while a client is offline, that client will not create conflict files for all those new files when it starts up.

1
newsfragments/2966.other
View File

@ -1 +0,0 @@
The NoNetworkGrid implementation has been somewhat improved.

1
newsfragments/2967.other
View File

@ -1 +0,0 @@
A bug in the test suite for the create-alias command has been fixed.

1
newsfragments/2968.other
View File

@ -1 +0,0 @@
The integration test suite has been updated to use pytest-twisted instead of deprecated pytest APIs.

1
newsfragments/2969.other
View File

@ -1 +0,0 @@
The magic-folder integration test suite now performs more aggressive cleanup of the processes it launches.

1
newsfragments/2970.other
View File

@ -1 +0,0 @@
The integration tests now correctly document the `--keep-tempdir` option.

1
newsfragments/2971.other
View File

@ -1 +0,0 @@
A misuse of super() in the integration tests has been fixed.

1
newsfragments/2972.feature
View File

@ -1 +0,0 @@
The Magic-Folder frontend now emits structured, causal logs. This makes it easier for developers to make sense of its behavior and for users to submit useful debugging information alongside problem reports.

1
newsfragments/2973.other
View File

@ -1 +0,0 @@
Several utilities to facilitate the use of the Eliot causal logging library have been introduced.

1
newsfragments/2974.other
View File

@ -1 +0,0 @@
The Windows CI configuration has been tweaked.

1
newsfragments/2975.feature
View File

@ -1 +0,0 @@
The `tahoe` CLI now accepts arguments for configuring structured logging messages which Tahoe-LAFS is being converted to emit. This change does not introduce any new defaults for on-filesystem logging.

1
newsfragments/2976.bugfix
View File

@ -1 +0,0 @@
The confusing and misplaced sub-command group headings in `tahoe --help` output have been removed.

1
newsfragments/2977.other
View File

@ -1 +0,0 @@
The Magic-Folder frontend has had additional logging improvements.

1
newsfragments/2997.bugfix
View File

@ -1 +0,0 @@
The Magic-Folder frontend is now more responsive to subtree changes on Windows.

1
newsfragments/3001.other
View File

@ -1 +0,0 @@
Added a simple sytax checker so that once a file has reached python3 compatibility, it will not regress.

1
newsfragments/3002.other
View File

@ -1 +0,0 @@
Converted all uses of the print statement to the print function in the ./misc/ directory.

1
newsfragments/3003.other
View File

@ -1 +0,0 @@
The contributor guidelines are now linked from the GitHub pull request creation page.

0
newsfragments/3005.minor
View File

1
newsfragments/3006.feature
View File

@ -1 +0,0 @@
The web API now publishes streaming Eliot logs via a token-protected WebSocket at /private/logs/v1.

0
newsfragments/3007.minor
View File

1
newsfragments/3008.other
View File

@ -1 +0,0 @@
Updated the testing code to use the print function instead of the print statement.

1
newsfragments/3009.other
View File

@ -1 +0,0 @@
Replaced print statement with print fuction for all tahoe_* scripts.

1
newsfragments/3010.other
View File

@ -1 +0,0 @@
Replaced all remaining instances of the print statement with the print function.

1
newsfragments/3011.other
View File

@ -1 +0,0 @@
Replace StringIO imports with six.moves.

0
newsfragments/3012.minor
View File

Some files were not shown because too many files have changed in this diff Show More