When a non-database exception is thrown out of a `withEntityManager`
block, always check if the session needs to be rolled back.
This means if a database error is caught and a new non-database error is
thrown out of the `withEntityManager` block, the transaction is still
rolled back. The flow can then continue progressing as normal.
* CORDA-3722 withEntityManager can rollback its session
Improve the handling of database transactions when using
`withEntityManager` inside a flow.
Extra changes have been included to improve the safety and
correctness of Corda around handling database transactions.
This focuses on allowing flows to catch errors that occur inside an
entity manager and handle them accordingly.
Errors can be caught in two places:
- Inside `withEntityManager`
- Outside `withEntityManager`
Further changes have been included to ensure that transactions are
rolled back correctly.
Errors caught inside `withEntityManager` require the flow to manually
`flush` the current session (the entity manager's individual session).
By manually flushing the session, a `try-catch` block can be placed
around the `flush` call, allowing possible exceptions to be caught.
Once an error is thrown from a call to `flush`, it is no longer possible
to use the same entity manager to trigger any database operations. The
only possible option is to rollback the changes from that session.
The flow can continue executing updates within the same session but they
will never be committed. What happens in this situation should be handled
by the flow. Explicitly restricting the scenario requires a lot of effort
and code. Instead, we should rely on the developer to control complex
workflows.
To continue updating the database after an error like this occurs, a new
`withEntityManager` block should be used (after catching the previous
error).
Exceptions can be caught around `withEntityManager` blocks. This allows
errors to be handled in the same way as stated above, except the need to
manually `flush` the session is removed. `withEntityManager` will
automatically `flush` a session if it has not been marked for rollback
due to an earlier error.
A `try-catch` can then be placed around the whole of the
`withEntityManager` block, allowing the error to be caught while not
committing any changes to the underlying database transaction.
To make `withEntityManager` blocks work like mini database transactions,
save points have been utilised. A new savepoint is created when opening
a `withEntityManager` block (along with a new session). It is then used
as a reference point to rollback to if the session errors and needs to
roll back. The savepoint is then released (independently from
completing successfully or failing).
Using save points means, that either all the statements inside the
entity manager are executed, or none of them are.
- A new session is created every time an entity manager is requested,
but this does not replace the flow's main underlying database session.
- `CordaPersistence.transaction` can now determine whether it needs
to execute its extra error handling code. This is needed to allow errors
escape `withEntityManager` blocks while allowing some of our exception
handling around subscribers (in `NodeVaultService`) to continue to work.
## Summary
This change deals with multiple issues:
* Errors that occur during flow initialisation.
* Errors that occur when handling the outcome of an existing flow error.
* Failures to rollback and close a database transaction when an error
occurs in `TransitionExecutorImpl`.
* Removal of create and commit transaction actions around retrying a flow.
## Errors that occur during flow initialisation
Flow initialisation has been moved into the try/catch that exists inside
`FlowStateMachineImpl.run`. This means if an error is thrown all the way
out of `initialiseFlow` (which should rarely happen) it will be caught
and move into a flow's standard error handling path. The flow should
then properly terminate.
`Event.Error` was changed to make the choice to rollback be optional.
Errors during flow initialisation cause the flow to not have a open
database transaction. Therefore there is no need to rollback.
## Errors that occur when handling the outcome of an existing flow error
When an error occurs a flow goes to the flow hospital and is given an
outcome event to address the original error. If the transition that was
processing the error outcome event (`StartErrorPropagation` and
`RetryFlowFromSafePoint`) has an error then the flow aborts and
nothing happens. This means that the flow is left in a runnable state.
To resolve this, we now retry the original error outcome event whenever
another error occurs doing so.
This is done by adding a new staff member that looks for
`ErrorStateTransitionException` thrown in the error code path of
`TransitionExecutorImpl`. It then takes the last outcome for that flow
and schedules it to run again. This scheduling runs with a backoff.
This means that a flow will continually retry the original error outcome
event until it completes it successfully.
## Failures to rollback and close a database transaction when an error occurs in `TransitionExecutorImpl`
Rolling back and closing the database transaction inside of
`TransitionExecutorImpl` is now done inside individual try/catch blocks
as this should not prevent the flow from continuing.
## Removal of create and commit transaction actions around retrying a flow
The database commit that occurs after retrying a flow can fail which
required some custom code just for that event to prevent inconsistent
behaviour. The transaction was only needed for reading checkpoints from
the database, therefore the transaction was moved into
`retryFlowFromSafePoint` instead and the commit removed.
If we need to commit data inside of `retryFlowFromSafePoint` in the
future, a commit should be added directly to `retryFlowFromSafePoint`.
The commit should occur before the flow is started on a new fiber.
The state machines state is held within `InnerState` which lived inside
the SMM. `InnerState` has been extracted out of the SMM to allow the SMM
to be refactored in the future. Smaller classes can now be made that
focus on a single goal as the locking of the state can be accessed from
external classes. To achieve this, pass the `InnerState` into the class
and request a lock if needed.
The locking of `InnerState` has been made a property of the `InnerState`
itself. It has a `lock` field that allows locks to be taken out when
needed.
An inline `withLock` function has been added to tidy up the code and not
harm performance.
Some classes have been made internal to prevent invalid usage of purely
node internal classes.
As part of this change, flow timeouts have been extracted out into
`FlowTimeoutScheduler`.
Only hit the database if `StateMachineState.isAnyCheckpointPersisted`
returns true. Otherwise, there will be no checkpoint to retrieve from the
database anyway. This can prevent errors due to a transient loss of
connection to the database.
Update tests after merging to 4.6
* Decouple DatabaseConfig and CordaPersistence etc.
* Add schema sync to schema migration + test
* Add command line parameters for synchronising schema
Only hit the database if `StateMachineState.isAnyCheckpointPersisted`
returns true. Otherwise, there will be no checkpoint to retrieve from the
database anyway. This can prevent errors due to a transient loss of
connection to the database.
* CORDA-3578 add corda prefix to conf file names as original issue was having non-corda reference.conf files on classpath causes DriverDSLImp failure
it is better to have this naming convention and avoid further conflicts of conf files.
* fixed weird duplicates
* revert renaming changes for web-reference.conf and loadtest-reference.conf
* Ensure logs directory is written to correct location
* Remove a superfluous set of log path property
* Add a unit test to catch bad log paths
* Address detekt issues
* checking for unsigned cordapps in prod mode and shutting down if it the contracts are not signed
* inheriting from CordaRuntimeException instead of Exception
* had the same tests twice, removed the shutdown for minimumplatformversion, modified some of the tests
* shut down when we get invalid platformversion and modified the test according to this
* making the message more accurate
* CORDA-3176 Inefficient query generated on vault queries with custom paging
a check added to avoid self joins
local postgres db tests were executed on large volume test data (50k states) to ensure that the DB optimizes the suspicious query and so it does not cuase performance issues.
A test added to check that a pagination query on large volume of sorted data is completed in a reasonable time
* foreach detekt fix
Adding to the query -explicitly- the flow's locked states resolved the SQL Deadlocks in SQL server. The SQL Deadlocks would come up from `softLockRelease` when only the `lockId` was passed in as argument. In that case the query optimizer would use `lock_id_idx(lock_id, state_status)` to search and update entries in `VAULT_STATES` table.
However, all rest of the queries would follow the opposite direction meaning they would use PK's `index(output_index, transaction_id)` but they would also update the `lock_id` column and therefore the `lock_id_idx` as well, because `lock_id` is a part of it. That was causing a circular locking among the different transactions (SQL processes) within the database.
To resolve this, whenever a flow attempts to reserve soft locks using their flow id (remember the flow id is always the flow id for the very first flow in the flow stack), we then save these states to the fiber. Then, upon releasing soft locks the fiber passes that set to the release soft locks query. That way the database query optimizer will use the primary key index of VAULT_STATES table, instead of lock_id_idx in order to search rows to update. That way the query will be aligned with the rest of the queries that are following that route as well (i.e. making use of the primary key), and therefore its locking order of resources within the database will be aligned with the rest queries' locking orders (solving SQL deadlocks).
* Fixed SQL deadlocks caused from softLockRelease, by saving locked states per fiber; NodeVaultService.softLockRelease query then uses VAULT_STATES PK index instead of lock_id_idx
Speed up SQL server by breaking down queries with > 16 elements in their IN clause, into sub queries with 16 elements max in their IN clauses
* Allow softLockedStates to remove states
* Add to softLockedStates only states soft locked under our flow id
* Fix softLockRelease not to take into account flowStateMachineImpl.softLockedStates when using lockId != ourFlowId
* Moved CriteriaBuilder.executeUpdate at the bottom of the file
Performance tuning of the new checkpoint schema;
- Checkpoint tables are now using `flowId` as join keys.
- Indexes consist of a PK's index on `node_checkpoints(flow_id)` and then unique indexes on `node_checkpoint_blobs(flow_id)` and `node_flow_metadata(flow_id)`.
- Serialization of `checkpointState` is being done with `CHECKPOINT_CONTEXT` so that we can have compression. This is needed when messages get passed into `checkpointState.sessions` therefore `checkpointState` grows in size upon serialized and
saved into the database.
* Deserialize checkpointState with CHECKPOINT_CONTEXT
* Align tests with schema update; We cannot add and update a checkpoint in the same session now, ends up with hibernate complaining: two different objects with same identifier
* Fix indentation and format
* Ignore tests that assert DBFlowResult or DBFlowException
* Set DBFlowCheckpoint.blob to null whenever the flow errors or hospitalizes; this way we save an extra SELECT in such cases;
* Fix test; cleared Hibernate session, it would fail at checkpoint with 'org.hibernate.NonUniqueObjectException'
* Changing VARCHAR to NVARCHAR
* Rename v17 liquibase scripts to v19 to resolve collision with ENT v17 scripts
* CORDA-3750: Use hand-written sandbox Crypto object that delegates to the node.
* CORDA-3750: Add integration test for deterministic CashIssueAndPayment flow.
* Tidy up generics for Array instances.
* Upgrade to DJVM 1.1-RC04.
When a flow is missing its database transaction the _real_ stacktrace
of the exception is missing in the logs. This is due to the normal error
handling path way does not work due the transaction being missing. When
it goes to process the next error event (due to the original transaction
context missing error) it will fail for the same error as it needs the
context to be able to process the error event.
The extra logging should help diagnose future errors.
Removing the ability to initialise schema from the node config, and add a new sub-command to initialise the schema (that does not do anything else and exits afterwards).
Also adding a command line flag that allow app schema to be maintained by hibernate for legacy cordapps, tests or rapid development.
Patching up mock net and driver test frameworks so they create the required schemas for tests to work, defaulting schema migration and hibernate schema management to true to match pre-existing behaviour.
Modified network bootstrapper to run an initial schema set-up so it can register nodes.
* Fix erroneous sql statement for oracle; It was failing tests with 'ORA-00933: SQL command not properly ended'
* Fixed flaky test; it didn't wait for counter party flow to get hospitalized as the test implied
Added command-line option: `--pause-all-flows` to the Node to control this.
This mode causes all checkpoints to be set to status PAUSED when the
state machine starts up (in StartMode.Safe mode).
Changed the state machine so that PAUSED checkpoints are loaded into
memory (the checkpoint is deserialised but the flow state is left serialised)
but not started.
Messages from peers are queued whilst the flow is paused and processed
once the flow is resumed.
When a non-database exception is thrown out of a `withEntityManager`
block, always check if the session needs to be rolled back.
This means if a database error is caught and a new non-database error is
thrown out of the `withEntityManager` block, the transaction is still
rolled back. The flow can then continue progressing as normal.
* CORDA-3715: When loading cordapps now check that contract classes have class version between 49 and 52
* CORDA-3715: Now check class version when contract verification takes place.
* CORDA-3715: Making detekt happy with number of levels in func
* CORDA-3715: Make use of new ClassGraph release which provides class file major version number.
* CORDA-3715: Changed package name in test jars
* CORDA-3715: Use ClassGraph when loading attachments.
* CORDA-3715: Reverted file to 4.5 version
* CORDA-3715: Updating method to match non deterministic version.
* CORDA-3715: Added in default param.
* CORDA-3715: Adjusted min JDK version to 1.1
* CORDA-3715: Switching check to JDK 1.2
* CORDA-3715: Now version check SerializationWhitelist classes.
* CORDA-3715: Switched default to null for range.
* [EG-438] First commit of error code interface
* [EG-438] Implement error reporter and a few error codes
* [EG-438] Add unit tests and default properties files
* [EG-438] Add the error table builder
* [EG-438] Update initial properties files
* [EG-438] Add some Irish tests and the build.gradle
* [EG-438] Fall back for aliases and use different resource strategy
* [EG-438] Define the URL using a project-specific context
* [EG-438] Tidy up initialization code
* [EG-438] Add testing to generator and tidy up
* [EG-438] Remove direct dependency on core and add own logging config
* [EG-438] Fix compiler warnings and tidy up logging
* [EG-438] Fix detekt warnings
* [EG-438] Improve error messages
* [EG-438] Address first set of review comments
* [EG-438] Use enums and a builder for the reporter
* [EG-438] Address first set of review comments
* [EG-438] Use enums and a builder for the reporter
* [EG-438] Add kdocs for error resource static methods
* [EG-440] Add error code for duplicate CorDapp loading
* [EG-438] Handle enums defined with underscores
* [EG-440] Add errors for some CorDapp loading scenarios
* [EG-440] Finish adding errors for CorDapp loading
* [EG-440] Fix up errors in properties files
* [EG-440] Start change to error code definition
* [EG-440] Update error code definition and add resource generation tool
* [EG-440] Tidy up error resource generation tool frontend
* [EG-440] Small refactorings and add kdocs
* [EG-440] Generate all missing resources
* [EG-440] Some refactoring and start writing a test
* [EG-440] Update unit test for resource generator
* [EG-440] Renaming of various parts of the error tool
* [EG-440] Add testing for errors and fix an issue in resource generation
* [EG-440] Add a kdoc for context provider API
* [EG-440] Remove old code from repository
* [EG-440] Address some review comments
* CORDA-3291 `isKilled` flag and session errors for killed flows
## Summary
Two major improvements have been worked on:
- A new flag named `isKilled` has been added to `FlowLogic` to allow
developers to break out of loops without suspension points.
- Killed flows now send session errors to their counter parties allowing
their flows to also terminate without further coordination.
Achieving these changes required a __fundamental__ change to how flows are
killed as well as how they sleep.
## `isKilled` flag
The addition of `FlowLogic.isKilled` allows flows to check if the
current flow has been killed. They can then throw an exception to lead
to the flow's termination (following the standard error pathway). They
can also perform some extra logic or not throw an exception if they
really wanted to.
No matter what, once the flag is set, the flow will terminate. Due to
timing, a killed flow might successfully process its next suspension
event, but it will then process a killed transition and terminate.
## Send session errors when killing a flow
A flow will now send session errors to all of its counter parties. They
are transferred as `UnexpectedFlowEndException`s. This allows initiated
flows to handle these errors as they see fit, although they should
probably just terminate.
## How flows are killed
### Before
Originally we were relying on Quasar to interrupt a flow's fiber, we
could then handle the resulting `InterruptedException`. The problem with
this solution is that it only worked when a flow was already suspended
or when a flow moved into suspension. Flows stuck in loops did not work.
### After
We now *do not* use Quasar to interrupt a flow's fiber. Instead, we
switch `FlowStateMachine.isKilled` to true and schedule a new event.
Any event that is processed after switching this flag will now cause a
`KilledFlowTransition`. This transition follows similar logic to how
error propagation works. Note, the extra event allows a suspended flow
to be killed without waiting for the event that it was _really_ waiting
for.
This allows a lot of the tidy up code in `StateMachineManager.killFlow`
to be removed as tidy up is executed as part of removing a flow.
Deleting a flow's checkpoint and releasing related soft locks is still
handled manually in case of infinite loops but also triggered as part
of the actions executed in a transition.
This required flow sleeping to be changed as we no longer rely on
quasar.
## How flows now sleep
The reliance on Quasar to make a flow sleep has been removed.
Instead, when a flow sleeps we create a `ScheduledFuture` that is
delayed for the requested sleep duration. When the future executes it
schedules a `WakeUpFromSleep` event that wakes up the flow... Duh.
`FlowSleepScheduler` handles the future logic. It also uses the same
scheduled thread pool that timed flows uses.
A future field was added to `StateMachineState`. This removes the
need for concurrency control around flow sleeps as the code path does
not need to touch any concurrent data structures.
To achieve this:
- `StateMachineState.future` added as a `var`
- When the `ScheduledFuture` is created to wake up the flow the passed
in `StateMachineState` has its `future` value changed
- When resumed `future` and `isWaitingForFuture` are set to `null` and
`false` respectively
- When cancelling a sleeping flow, the `future` is cancelled and nulled
out. `isWaitingForFuture` is not changed since the flow is ending anyway
so really the value of the field is not important.
* [EG-438] First commit of error code interface
* [EG-438] Implement error reporter and a few error codes
* [EG-438] Add unit tests and default properties files
* [EG-438] Add the error table builder
* [EG-438] Update initial properties files
* [EG-438] Add some Irish tests and the build.gradle
* [EG-438] Fall back for aliases and use different resource strategy
* [EG-438] Define the URL using a project-specific context
* [EG-438] Tidy up initialization code
* [EG-438] Add testing to generator and tidy up
* [EG-438] Remove direct dependency on core and add own logging config
* [EG-438] Fix compiler warnings and tidy up logging
* [EG-438] Fix detekt warnings
* [EG-438] Improve error messages
* [EG-438] Address first set of review comments
* [EG-438] Use enums and a builder for the reporter
* [EG-438] Add kdocs for error resource static methods
* [EG-438] Handle enums defined with underscores
* [EG-438] Slight refactoring of startup code
* [EG-438] Port changes to error reporting code from future branch
* [EG-438] Also port test changes
* [EG-438] Suppress a deliberately unused parameter
* CORDA-3722 withEntityManager can rollback its session
## Summary
Improve the handling of database transactions when using
`withEntityManager` inside a flow.
Extra changes have been included to improve the safety and
correctness of Corda around handling database transactions.
This focuses on allowing flows to catch errors that occur inside an
entity manager and handle them accordingly.
Errors can be caught in two places:
- Inside `withEntityManager`
- Outside `withEntityManager`
Further changes have been included to ensure that transactions are
rolled back correctly.
## Catching errors inside `withEntityManager`
Errors caught inside `withEntityManager` require the flow to manually
`flush` the current session (the entity manager's individual session).
By manually flushing the session, a `try-catch` block can be placed
around the `flush` call, allowing possible exceptions to be caught.
Once an error is thrown from a call to `flush`, it is no longer possible
to use the same entity manager to trigger any database operations. The
only possible option is to rollback the changes from that session.
The flow can continue executing updates within the same session but they
will never be committed. What happens in this situation should be handled
by the flow. Explicitly restricting the scenario requires a lot of effort
and code. Instead, we should rely on the developer to control complex
workflows.
To continue updating the database after an error like this occurs, a new
`withEntityManager` block should be used (after catching the previous
error).
## Catching errors outside `withEntityManager`
Exceptions can be caught around `withEntityManager` blocks. This allows
errors to be handled in the same way as stated above, except the need to
manually `flush` the session is removed. `withEntityManager` will
automatically `flush` a session if it has not been marked for rollback
due to an earlier error.
A `try-catch` can then be placed around the whole of the
`withEntityManager` block, allowing the error to be caught while not
committing any changes to the underlying database transaction.
## Savepoints / Transactionality
To make `withEntityManager` blocks work like mini database transactions,
save points have been utilised. A new savepoint is created when opening
a `withEntityManager` block (along with a new session). It is then used
as a reference point to rollback to if the session errors and needs to
roll back. The savepoint is then released (independently from
completing successfully or failing).
Using save points means, that either all the statements inside the
entity manager are executed, or none of them are.
## Some implementation details
- A new session is created every time an entity manager is requested,
but this does not replace the flow's main underlying database session.
- `CordaPersistence.transaction` can now determine whether it needs
to execute its extra error handling code. This is needed to allow errors
escape `withEntityManager` blocks while allowing some of our exception
handling around subscribers (in `NodeVaultService`) to continue to work.
On node start, load CordaServices before starting the NotaryService,
so that the NotaryService can check that the services it requires are
available when starting.
Resolves#6172.
* CORDA-3762: Integration test exposing the problem reported
* CORDA-3726: Additional logging
* CORDA-3726: Prevent thread leaks
* CORDA-3726: New `journalBufferTimeout` parameter
* CORDA-3726: Override `journalBufferTimeout` parameter
* CORDA-3726: Making Detekt happier
* CORDA-3276: Account for extra thread user in MockNetwork
For real node this does not matter as `shutdown` can safely be called multiple times, which is not true for server thread provided by MockNetwork
* CORDA-3276: Do not make SMM shutdown "executor" as it belongs to AbstractNode
* CORDA-3276: Address input from @rick-r3
* CORDA-3276: Fix test after rebase
* adding blocked functions ro RestrictedEntityManager and creating RestrictedConnection class
* adding flow tests and fixing issues regarding the review
* adding quasar util to gradle
* updating flow tests
* adding space before } at .isThrownBy()
* adding spaces
* [EG-503] Spent state audit tool
Fixes
* Refinements to notary query interfaces. Feature complete.
* EG-503: Introduce optional `notaryService` in `ServiceHubCoreInternal`
* Remove redundant logic following change to use extensions API
Co-authored-by: Viktor Kolomeyko <viktor.kolomeyko@r3.com>
* CORDA-3696: Temporary update to enable JDK11 build and test. Will eventually be switchable.
* CORDA-3696: Filter out the Nashorn warning.
* CORDA-3696: Add JDK11 classifier.
* CORDA-3696: Updated match string to cope with JDK11.
* CORDA-3696: Filtering out SPHINCS256_SHA256 where failing due to JDK11.
* CORDA-3696: Now remove SPHINCS256_SHA256 only if JDK11.
* CORDA-3696: Fix test failure - switch to regex matching.
* CORDA-3696: Hide the illegal access warnings.
* CORDA-3696: Check for Java11 when disabling Java11 warnings.
* CORDA-3696: Fix unneccessary non null check.
* CORDA-3696: Reverting build env to JDK8
* CORDA-3696: Revert hiding of illegal access warnings via Unsafe class.
* CORDA-3696: Remove internal access warnings and new JDK11 version checker.
* CORDA-3696: Updated build file for OS
* CORDA-3696: Removed typo
* CORDA-3696: Fixed space typo.
* CORDA-3696: Open modules to remove the illegal access warnings.
Co-authored-by: Adel El-Beik <adelel-beik@19LDN-MAC108.local>
* CORDA-3691 Delete checkpoint when flow finishes
The checkpoint and its related records in joined tables should be deleted
when a flow finishes.
Keeping these flows around will be completed in the future.
* CORDA-3691 Ignore some flow metadata tests
Ignore tests around recording the finish time of flow metadata records
since we are not currently keeping COMPLETED flows in the database.
Flows that are kept for overnight observation:
- Save their Checkpoint.status as 'HOSPITALIZED' in the database
- Save the error that caused the hospitalization in the database
A new Event was added for this reason. Whenever the hospital determines
a flow for hospitalization, it adds this Event in the flow's fiber queue.
When processed it creates a new DB transaction, stores the checkpoint status along with
the error, and it adds a 'FlowContinuation.ProcessEvents' continuation so that the fiber keeps
processing events (effectively since there are no more events in the fiber's channel, the fiber will suspend).
Flows that error:
- Their checkpoints are kept in the database
- Save their Checkpoint.status as 'FAILED'
- Save the error that caused the error in the database
Upon erroring, the flow's Checkpoint.status gets updated('FAILED') and the checkpoint is stored
in the database instead of getting removed. The flow then propagates the error to counterparties,
sets its future with the error and gets removed from memory.
* ENT-4967: Require no classifier for corda-node-djvm, corda-deserializers-djvm.
* Also remove classifiers from core, serialization and finance-contracts.
* Compile corda-serialization-djvm for Java 8 and remove its classifier.
Added a new field Completed to the in-memory object FlowState.
FlowState.Completed is corresponds to flow_state=Null in the DB.
This change will save disk space.
* Run serialisation tests with both in-process and out-of-process nodes.
* Add custom serialisers and whitelists to Driver's AMQPServerSerializationScheme.
* Run serialisation tests with both in-process and out-of-process nodes.
* Add custom serialisers and whitelists to Driver's AMQPServerSerializationScheme.
* CORDA-3601 Record a flow's finish time
Record a flow's finish time by updating its metadata record. It is set
in `updateCheckpoint` by checking the status of the checkpoint. If it is
`COMPLETED` it will set the `finishInstant` on the metadata object and
update it.
* CORDA-3601 Record flow finish time for all finished statuses
Update the flow finish time for the following statuses:
- COMPLETED
- KILLED
- FAILED
* CORDA-3601 Use platform clock in `DBCheckpointStorage`
* CORDA-3669 Do not execute `ExecuteAsyncOperation` multiple times
When a `FlowExternalOperation` or `FlowExternalAsyncOperation` executes
and completes a flag (`isFlowResumed`) is switched to true.
This flag was used inside of `DoRemainingWorkTransition` to decide
whether to skip over the execution of an event.
Since this flag was being switched to true when the external operation's
future completed, it was possible for _unexpected_ events to be placed
in the fiber's queue that would retrigger the
`FlowIORequest.ExecuteAsyncOperation`, that is held as the checkpoint's
next `FlowIORequest`to process.
By using the existing `StateMachineState.isTransactionTracked` (and
renaming it to `isWaitingForFuture`) we can decide to not process the
`FlowIORequest.ExecuteAsyncOperation` if it has already been called
before. This moves this code path in line with
`FlowIORequest.WaitForLedgerCommit`.
Random `DoRemainingWork` events can now be pushed to the fiber's queue
without causing the `FlowIORequest.ExecuteAsyncOperation` to execute
again.
* CORDA-3596 Record flow metadata
Record flow metadata during the zero'th checkpoint that occurs before
calling the flow's `call` function.
This required adding an RPC call's arguments to the `InvocationContext`
that gets created. These arguments are then accessible within the
statemachine and from the `Checkpoint` class. The arguments are then
extracted when recording a flow's metadata inside of
`DBCheckpointStorage`.
Updated the size of the started by column to 128 since it was not long
enough to hold the fully qualified class of a service that started a
flow.
* CORDA-3596 Remove arguments from in-memory checkpoint
When executing a flows first real suspend (from flow code) the arguments
contained in the `InvocationContext` are removed. This saves holding
these arguments for the whole lifecyle of a flow.
* CORDA-3596 Increase `cordapp_name` column to 128
* CORDA-3596 Join metadata by `flow_id`
Due to changes in where metadata is recorded, there is no need for
having `invocation_id` as the metadata table's primary key. The
`flow_id` is now the primary key of the table and is used to join to the
main checkpoints table.
The `invocation_id` has been removed from the checkpoints table since it
is not needed for the join anymore.
* CORDA-3596 Remove `received_time` from metadata table
* CORDA-3596 Remove unused `StartReason` enum
* CORDA-3596 Simple `DBCheckpointStorageTests` for metadata
* CORDA-3596 Truncate really long flow names
* CheckpointStorage.getAllCheckpoints will not fetch COMPLETED, FAILED and KILLED flows by default
* Rename getAllCheckpoints to getAllRunnableCheckpoints for clarity
* Fix Detekt issue
* Rename getAllRunnableCheckpoints to getRunnableCheckpoints
* Minor kdoc update
* Bring back in CheckpointStorage.getAllCheckpoints to co-exist with getRunnableCheckpoints
* Add progress tracker information to checkpoint
The checkpoint Datebase is updated when the statemachine suspends
with the progress trackers current step name. This is truncated if
it is longer than the Database column.
* Minor rename in statemachine for clarity
* Set/ Reset Checkpoint.status to RUNNABLE after when suspending
* Removing/ Moving comment as it makes no longer sense to be there since, we now always create a new Checkpoint object in SingleThreadedStateMachineManager.createFlowFromCheckpoint through tryDeserializeCheckpoint
* Set -in memory- Checkpoint.status to RUNNABLE when a flow is retrying from Checkpoint
Due to a change in how messaging works, `ActionExecutorImpl
.executeSendInitial` was no longer being called. Changing the byteman
script to throw exception on hits to `ActionExecutorImpl
.executeSendMultiple` allowed the tests to pass.
When a flow is finished do not delete the checkpoint from the DB.
Instead, the FlowStatus is marked as Completed in the DB.
Updated numerous tests which relied on the flow being removed
when finished.
* Update Checkpoint DB to update flow io request
* Modify flow monitor to update Checkpoint DB with waiting flows
This happens periodically.
* Refactored code to avoid looping twice and updated tests
* Fix tests after rebasing
* Fix MR comments (non-functional refactor of tests + FlowMonitor).
* Made visible for testing method private in DBCheckpointStorage
This is not needed anymore.
* Explicity check if ioRequestType has changed in update method
* Fix shadowing warning
* Import non deprecated Assert into test
* Use AssertEquals not assert in test
* Address more comments (minor refactor) of DBCheckpointStorage
* Minor fix use it instead of referencing object explicitly
* Add null check to DBCheckpointStorage
* Revert changes to Flow Monitor.
We will instead store the information in the main thread of the
state machine.
* Remove now uneeded API and make statemachine update ioRequest
* Add Integration Test to check statemachine updates DB on Recieve
* Use simpleName in checkpoint storage instead of class.
Hibernate was previously resetting the class field this is now
set to null (when getting checkpoint form DB) and a new method
for getting back the simple name as a string.
* Update StateMachineState to store simple name.
* Fix after rebase broke stuff + renamed test
* Fix Detekt issue
* Remove uneeded null assertion
* [CORDA-3628] - Implement sendAll API
* detekt
* Some minor refactorings and docs
* Eliminate warnings
* Address Rick's comments
* Switch sendAll to use a set
Do not cascade updates to checkpoint error and result tables to hopefully
improve database performance moving forward. Because the joined tables
are no longer being updated by updating the main `DBFlowCheckpoint` entity,
they must be created/updated/deleted manually.
The checkpoint blobs still cascade as they pretty much always evolve in
tandem with the main checkpoint table.
* CORDA-3651: addManifest now uses separate files for reading and writing.
* CORDA-3651: The jar scanning loader now closes itsself.
Co-authored-by: Adel El-Beik <adelel-beik@19LDN-MAC108.local>
* Split out node-api tests that require test-utils/node-driver
* Add node-api test artefacts to publication list.
* Make test-common a transient dependency - downstream tests assume that it's available.
* Switch dependencies to java-library
* Fix magic package name for cordapp scanning in test
* CORDA-3644: Scan the CorDapp classloader directly for SerializationWhitelist.
* CORDA-3644: Filter CorDapps from out-of-process node classpaths by their manifest attributes. Also exclude directories and blatant test artifacts.
* Fix IRS Demo - its "tests" artifact had a non-standard classifier of "test".
* Remove unused dependencies from test-common
* Explicit imports and formatting
* Add core-test-utils project
* Add dependency
* Move Kryo serialization context to node-api (not serialization as we do not want to pull kryo into the serialization lib)
* Move AMQP server serialization scheme to node api
* Move serialization tests to node-api
* Move internal test helpers without further dependencies.
* Move out some types from RPCClientProxyHandler to node-api in preparation for moving the AMQP scheme
* Move client AMQP context to node-api so we can move the test serialization rule out.
* Move InternalSerializationTestHelpers to core-test-utils
* Moved testing.core to core-test-utils
* Make detekt happy
* Add api-scanner to core-test-utils
* Remove inlined package names introduced by IntelliJ refactoring
* Update api-current.txt to account for reordering.
* Add core-test-utils to list of published artifacts.
* Add missing import
* Location of things in api text has moved again (publish name of artefact?)
* Revert all additions to the API, leaving just the reordering
* Code review: fix up core-test-utils build.gradle and introduce kryo version constant.
* Remove OpenSsl flag from ssl config stub (can't be used from node-api)
* Suppress detekt warning
* Move core test util tests to the right module
* Expose kotlin test as a transient dependency - projects have come to rely on that.
* Fix typo in package name
* Replace old Checkpoint table with new one.
Adds some of the new fields into the table where needed (I have
guessed this stuff but we can update it as we go along).
* Fix database constraints + name table correctly opps.
* Fixed typos in Liquidbase script
Also corrected constraints and added missed fields in hibernate
checkpoint class and liquibase scripts.
* Update CheckpointStorage to pass in serialization context.
This is cleaner than passing both the checkpoint and the
serialized checkpoint into the methods. Also fixed
CordaPersistanceServiceTests which I accidentally broke.
* Fix detekt problem
* Revert "Update CheckpointStorage to pass in serialization context."
This reverts commit b71e78f202.
* Fix test broken by reverting commit
* CORDA-3597 Update metadata join, timestamp columns and serialization
- Change the metadata join to the checkpoints table to use
`invocation_id` instead of `flow_id`. There were issues joining
between the tables because `flow_id` was not the primary key of the
metadata table. Switching over to `invocation_id` has at least allowed
us to bypass this issue. The information about the `invocation_id` is
stored in the `Checkpoint` class which makes it simple to save at
runtime.
- Some of timestamp columns were nullable when they should always be
populated, the nullable flags have now been removed.
- Previously the whole checkpoint was being serialized and stored into
the `checkpoints_blob.checkpoint` column. This meant duplicated saving
as the `flow_state` was contained in this object. Only the
`CheckpointState` property of `Checkpoint` is now being serialized and
saved to this field. Furthermore, it now uses the default
`STORAGE_CONTEXT` serialization (AMQP) instead of Kryo (which is only
used for serializing the `flow_state` / flow stack).
- The checkpoint database performance metrics recording has been
abstracted to its own class.
* CORDA-3597 Make metadata join non optional
Remove the nullable declaration on the metadata field of
`DBFlowCheckpoint`
* CORDA-3597 Rename `node_checkpoints_blobs` to `node_checkpoint_blobs`
* CORDA-3597 Update some kdocs
Co-authored-by: Dan Newton <danknewton@hotmail.com>
* CORDA-3484: Now cope with 2 contract jars with same hash but different name, we just select one and use that.
* ENT-3584: Contract jars are now generated on the fly.
* CORDA-3584: Reverted changes to CordappProviderImpl. Exception is raised if node started with multiple jars with same hash.
* ENT-3584: Fixing test failure.
* CORDA-3584: Switch to test extension method instead of reflection to access internal member.
* ENT-3584: Address review comment. Dont fully qualify exception.
* CORDA-3584: Address review comment and converted lazy to a resettable one.
* CORDA-3584: Removed unused logger.
* CORDA-3584: Fixed visibility.
* CORDA-3584: Removed synchronized
* CORDA-3584: Removed CordappResolver
* CORDA-3584: Reverted change in gradle file and fixed test.
* CORDA-3584: Removed V3 from test description as it wasn't actually V3 specific.
* CORDA-3584: Address review comment. Let classes be garbage collected.
* Throw SQLException or PersistenceException plain, that may come out of an unsafe subscriber
* Add explanatory comment about why we changed Observer.tee to use unsafe subscribe
* Introducing not unsubscribing version of Rx.Subscriber
* Wrap PublishSubjects with FlowSafeSubjects in all tests that test Observer.tee
* Minor code formatting
* Make rawUpdates Rx.Observers not unsubscribe when accessed from CordaServices - Do not allow rawUpdates subscribing from flows
* Warning fix: Add else block to when statement
* Revert "Wrap PublishSubjects with FlowSafeSubjects in all tests that test Observer.tee"
This reverts commit e419af86
* Correcting log message
* Improve log message
* Add fiber's id to log message and exception message
* Added test, asserting FlowSafeSubscriber is alive and re-accessed upon flow retry
* Logging flow name instead of flow id at VaultService.rawUpdates subscribing error
* Add kdoc to OnNextFailedException
* Minor text correction
* Update kdocs of FlowSafeSubject/ PreventSubscriptionsSubject
* Moved FlowSafeSubject under package node.internal as it is only used by NodeVaultService
* Add comment and update kdoc explaining how to subscribe with SafeSubscriber to FlowSafeSubject
* Change PreventSubscriptionsSubject#errorAction to be more specific; to return an Exception
* Minor text update
* Update messy comment
* Replace assertThat with assertEquals
* Splitting heartBeat to heartBeat1 and hearBeat2 for more clear asserting
* Correcting comment
* Update messy comment
* Splitting heartBeat into heartBeatOnNext and heartBeatOnError
* Update test name
* Add explanatory comment to test
* Update test name
* Update test and add test comment
* Moving NotarisedTxs from SendStateFlow to VaultObserverExceptionTest inside NodeHandle.getNotarisedTransactionIds
* Moving SubscribingRawUpdatesFlow from ErrorHandling to VaultObserverExceptionTest
* Update kdoc of FlowSafeSubscriber and FlowSafeSubscriber.onNext
* Make kdoc more clear
* Throw exception upon accessing VaultService.rawUpdates from within a flow
* Changing exception thrown when accessing VaultService.rawUpdates from within a flow to a CordaRuntimeException
* Minor kdoc update
* Update test comment
* Update kdoc of FlowSafeSubscriber
* Introducing Observable.flowSafeSubscribe public API method to subscribe with -non unsubscribing- Rx.Subscribers to Observables. It also replaced FlowSafeSubject
* Move CustomSafeSubscriber outside test methods
* Minor text update
* Add timeout to tests
* Update kdoc of flowSafeSubscribe
* Update kdoc of flowSafeSubscribe
* Update kdoc of flowSafeSubscribe
* Move FlowSafeSubscriber and flowSafeSubscribe under their own package
* Fix detekt issue
* Update Detekt baseline
* Revert "Update Detekt baseline"
This reverts commit 793a8ed9
* Fix Detekt issue
* Moved strictMode flag from flowSafeSubscribe to OnFlowSafeSubscribe
Moved OnFlowSafeSubscribe into internal package
Integration tested flowSafeLooseSubscribe
* Suppress Rx Deprecation
* Rename flowSafeSubscribe to flowSafeObservable
* Renaming flowSafeObservable to continueOnError and FlowSafeSubscriber to ResilientSubscriber
* Split StateMachine State into 2 classes
The idea is this better reflects the database structure. Added a
few helper methods to copy and update state.
* Doc + Improve Checkpoint API
* Rename methods to be more clear
* [NOTICK] Add a custom detekt rule for tests with no timeout, and fix remaining missing timeouts
* [NOTICK] Add a test for custom detekt rules and tidying
* add timeout annotation to new test
Co-authored-by: Stefano Franz <roastario@gmail.com>
* [EG-140] Allow system property paths with multiple keys to be specified in node.conf
* [EG-140] Split property paths to remove quotes
* [EG-140] Quote system properties in docs
* [EG-140] Rename path to key
* TM-197 Setting bouncy castle provider in order for the test to pass
* TM-197 setting timeout for all builds at 3 hours
* TM-197 ignoring unstable tests
* TM-197 switching 4.3 to use local k8s instances and also make the maximum duration of builds 3 hours, fix 1 test and ignore 2 flaky ones
* update to use local-k8s version of the plugin
Co-authored-by: Stefano Franz <roastario@gmail.com>
* CORDA-3565: `ServiceStateSupport` and supporting classes
* CORDA-3565:Plug `ServiceLifecycleSupport` into `MessagingService`
* CORDA-3565: Detekt baseline update
* CORDA-3565: React to MessagingServer going up and addition logging for up/down
Co-authored-by: Matthew Nesbit <matthew.nesbit@r3.com>
* Make tee not wrap PublishSubjects in SafeSubscribers, otherwise a non Rx exception from an unsafe observer shuts down all other observers under the same PublishSubject
* Throw SQLException or PersistenceException plain, that may come out of an unsafe subscriber
* Revert "Throw SQLException or PersistenceException plain, that may come out of an unsafe subscriber"
This reverts commit c7b8af3fa6.
* Update Detekt baseline
* Just passing in the exception was printing the entire stacktrace to the log, now we just pring the exception and message
* Updating exception message
Observers registered on NodeVaultService#rawUpdates, if they throw an exception when called from serviceHub#recordTransactions and if this exception is not handled by the flow hospital, then this leads to the transaction not being recorded in the local vault. This could get the ledger in an out of sync state.
In the specific case this happens within FinalityFlow#notariseAndRecord this leads to the transaction being notarized but not recorded in the local vault nor broadcasted in any counter party. The -failed to be recorded locally- transaction and its output states are not visible to any vault, and its input states not able to consumed by a new transaction, since they are recorded as consumed within the Notary. In this specific case we need not loose, by any means, the current transaction.
We will handle all cases by catching all exceptions thrown from serviceHub#recordTransactions, wrapping them with a HospitalizeFlowException and throwing it instead. The flow will get to the hospital for observation to be retried from previous checkpoint on next node restart.
Use `flowId` from `ExternalMessageEvent` when failing to init sessions instead of generating
a new random UUID. The a `flowId` is generated and stored inside the event after the
state machine work that done previously.
Cap the default size of the external operation thread pool to 10 or
the maximum number of available processors, whichever is smaller.
Set the minimum size of the thread pool to 1. Meaning that only a
single thread is used unless the node actually starts to use
`FlowExternalOperation` which consumes threads from this pool.
* CORDA-2942: Allow exception from `CordaService` creation to propagate
It will ultimately be thrown from Node's `start()` method terminating the node start-up sequence.
* CORDA-2942: Be lenient when retrievign the name of the Notary
Some tests setup such that they do nto have Notary running.
* CORDA-3549: Improve stability of `CordaServiceLifecycleFatalTests`
* CORDA-3549: Bump-up reps count to ensure that test is definitely not flaky when executed by CI
(once proved the number of reps will be reduced)
* CORDA-3549: Making Detekt happier
* CORDA-2942: Ensure `NodeLifecycleEventsDistributor` cleans-up smoothly when node shuts down
Deprecate FlowAsyncOperation and reimplement public versions FlowExternalOperation and FlowExternalAsyncOperation.
await added to FlowLogic to allow easy calling from both Java and Kotlin. There are two overrides of await (one for FlowExternalOperation and FlowExternalAsyncOperation).
Implementations of FlowExternalOperation return a result (written as blocking code) from their execute function. This operation will then be executed using a thread provided by the externalOperationExecutor.
Implementations of FlowExternalAsyncOperation return a future from their execute function. This operation must be executed on a newly spawned thread or one provided by a thread pool. It is up to developers to handle threading in this scenario.
The default thread pool (externalOperationExecutor) can be configured through the flowExternalOperationThreadPoolSize node config.
The current implementation leaves FlowAsyncOperation alone, meaning that any developers that have used it (even though it is internal) won't need to change their apps. If this was not concern I would delete it completely and replumb the state machine code. Instead, it has been marked with @DoNotImplement and executeAsync is annotated with @Deprecated
* CORDA-2942: Port minimal set of changes to make lifecycle events work
... and make codebase compile.
* CORDA-2942: Undo some changes which are not strictly speaking necessary
* CORDA-2942: Make `NodeServicesContext` leaner and delete `extensions-api` module
* CORDA-2942: Reduce even more number of files affected
* CORDA-2942: Integration test fix
* CORDA-2942: Make events `AfterStart` and `BeforeStop` generic w.r.t. `NodeServicesContext`
* CORDA-2942: `NodeLifecycleObserverService` and a set of integration tests.
Public API violations are expected as well as integration tests failing.
* CORDA-2942: Re-work to introduce `ServiceLifecycleObserver`
* CORDA-2942: Explicitly mention a type of exception that may be thrown for some events.
* CORDA-2942: Register `ServiceLifecycleObserver` through `AppServiceHub`
* CORDA-2942: Fix integration test + KDocs update
* CORDA-2942: Detekt and `api-current` update
* CORDA-2942: Improvement to `CordaServiceLifecycleFatalTests`
... or else it has side effects on other tests.
* CORDA-2942: Add an integration test for new API use in Java
Driver test is written in Kotlin, but services definition is written in Java.
Also KDocs improvements.
* CORDA-2942: Documentation and release notes update
* CORDA-2942: First set of changes following review by @mnesbit
* CORDA-2942: Second set of changes following review by @mnesbit
* CORDA-2942: Added multi-threaded test
* CORDA-2942: Fixes
* CORDA-2942: Undo changes to `api-current.txt`
* CORDA-2942: Bare mimimum change to `api-current.txt` for CI gate to pass.
* CORDA-2942: Address review feedback from @rick-r3
* CORDA-2942: Detekt update
* CORDA-2942: Delete `ServiceLifecycleObserverPriority` and replace it with `Int` after discussion with @mnesbit
* CORDA-2942: Introduce more `NodeLifecycleEvent` and switch services to listen for those events
* CORDA-2942: Few more changes after input from @rick-r3
* First stub on integration test
Unfinished - hang on issue and pay
* CORDA-2942: Switch to use out-of-process nodes for the inetgration test
Currently Alice and Notary stuck waiting to hear from each other.
* CORDA-2942: Extra log lines during event distribution
* CORDA-2942: Asynchronously distribute lifecycle events
* CORDA-2942: Await for complete P2P client start-up
Next step: Add vault query to integration test
* CORDA-2942: Asynchronously distribute lifecycle events
Next step: Improve integration test
* CORDA-2942: Fix test broken by recent changes and improve logging
* CORDA-2942: Improvement of the test to be able to monitor actions performed by @CordaService in the remote process
* CORDA-2942: Add node re-start step to the integration test
* CORDA-2942: Remove `CORDAPP_STOPPED` event for now
* CORDA-2942: s/CORDAPP_STARTED/STATE_MACHINE_STARTED/
* CORDA-2942: Inverse the meaning of `priority` as requested by @rick-r3
* CORDA-2942: Register `AppServiceHubImpl` for lifecycle events and put a warning when SMM is not ready.
* Do not register cordapp custom serialisers when using attachment classloader.
* Record the URLs of CorDapp JARs that contain custom serialisers. Include these JARs as extra attachments if we discover that we're missing a custom serialiser during transaction verification.
* Check for disabled serializer when explicitly requesting a custom serializer.
Refactor test case to force use of a custom serializer.
* Tidy up basic custom serializer test.
* Also test that TransactionBuilder rejects missing custom serializers.
* Remove test whitelists, which should not be needed with custom serialisers.
* Add changelog entry. Also align TestCordappImpl.findRoots() with OS backports.
* Second approach based around CorDapps inside AttachmentStorage - report missing type descriptor or any non-composable types.
* Initial implementation of Corda-Fixup rules inside a CorDapp jar.
* Replace original "automatic attachment fixing" mechanism completely.
* First review comments: restore "missing class" logic to TransactionBuilder.
* Restore "missing class" mechanism as fallback for SignedTransaction too.
* CORDA-3507: Use the config value for connectionRetryInterval rather than a hardcoded value
* CORDA-3507: Use the config value for connectionRetryInterval rather than a hardcoded value
* CORDA-3452: Node: Configure the input of custom string in CSR to be used by Identity Service
* CORDA-3452: Remove unused import
* CORDA-3452: Add test for networkServices configuration
* [CORDA-3436] Allow CorDapps access to node diagnostic information
* [CORDA-3436] Fix API breakages
* [CORDA-3436] Improve documentation around diagnostics service
* [CORDA-3436] Remove CorDapps from the diagnostics information
* [CORDA-3436] Silence detekt warning
* CORDA-3513: Don't try to reconnect for PermissionExceptions
* CORDA-3513: Don't try to reconnect for PermissionExceptions
* CORDA-3513: Add test for not reconnecting for PermissionExceptions
* CORDA-3513: Update exception message and test
* CORDA-2942: Switch to use predictable timestamp
* CORDA-2942: Validate content of dumped checkpoint
* CORDA-2942: First stub on the integration test
(no checkpoints dumped for some reason using RPC)
* CORDA-2942: Reduce checkpointing code to bare minimum
* CORDA-2942: Minor refactoring
* CORDA-2942: Verify dump checkpoint content
* ENT-4382: Move `InvocationHandlerTemplate` into `core`
This is an internal helper which is general enough and does not have any Node specific code.
* ENT-4382: Make @CordaInternal applicable to classes
And apply it on `AttachmentTrustCalculator` which is `core/internal` interface anyway.
* ENT-4237: Added timestamp to the node_transactions table.
* ENT-4237: Clock for timestamp now retrieved from ServiceHub. And now record verification time as well.
* ENT-4237: Fixed tests. Also enabled stream output in allParallelIntegrationTest.
* ENT-4237: Changed timestamp to a val.
* ENT-4237: Changed streamOutput to false for allParallelIntegrationTest
* ENT-4237: Unit tests added for new timestamp column. Also now passing a clock into DBTransactionStorage.
* ENT-4237: Added more unit tests to check timestamp
* ENT-4237: Fix test to actually change clock time when testing transaction time does not change.
* Introducing a new type of exception and a new hospital staff member to pause flows by immediately hospitalising them.
* Renaming exception to "HospitalizeFlowException".
* Making HospitalizeFlowException an open class.
* Overloading constructors of HospitalizeFlowException to be available in Java.
* Using Throwable#mentionsThrowable.
* Moving HospitalizeFlowException in its own file.
* Update kdocs for HospitalizeFlowException and StaffedFlowHospital#SedationNurse.
* Added tests, testing various HospitalizeFlowException types thrown.
* Fix Detekt issues.
* Imports optimizing.
* Add safe casting.
* Update api-flows and node-flow-hospital docs.
* Minor code comment change.
* Add DOCSTART-DOCEND signs in HospitalizeFlowException for makeDocs. It is referenced by api-flows.rst.
* Minor change in note.
* Code formatting.
* Remove comment.
* Remove if statement that makes example worse.
* Remove redundant comment.
* Moving 'Internal Corda errors' at the bottom.
* Changing node-flow-hospital.rst as per review.
* Change HospitalizeFlowException description as per review.
* Adding an example for FlowException.
* Minor indentation fix.
* Update FlowException example label as per review.
* Correcting handling of custom exception.
* Harmonize serialization/core and deterministic counterparts
* Fix test for changed private alias key behaviour
* Detekt errors
* roll back project.xml
* CORDA-3471: Create `CordaTransactionSupport` and use wherever possible instead of `CordaPersistence`
* CORDA-3471: Address comments by @mnesbit
- Relocate `CordaTransactionSupport` to `core`
- Create a lighter version of transaction - `VaultTransaction` that gives access to `session` object only.
* CORDA-3471: More changes after discussion with @mnesbit
- Rename `VaultTransaction` into `SessionScope`.
* CORDA-3471: Revert changes to most of the files after conversation with @mnesbit and @rick-r3
* CORDA-3471: Introduce `CordaTransactionSupportImpl` and make it accessible via `AppServiceHub`.
* CORDA-3471: Minor change (comment).
* CORDA-3471: Address input from @mnesbit
* CORDA-3471: Address input from @rick-r3
* CORDA-3471: Make Detekt happier
* CORDA-3471: Add a new test that proves transactions can be started from client threads
As requested by @mnesbit
* CORDA-3471: Change log and documentation update.
As requested by @mnesbit
* CORDA-3464: Also scan attachment:// URLs for custom serializers.
* Only scan the given classloader - ignore this classloader's parents.
* Upgrade to ClassGraph 4.8.58 - for "robustness fixes".
* Register the attachment:// URL scheme using AttachmentsClassLoader.
* Add integration test for custom serializer in contract state.
* Rename Currancy -> Currantsy, just to make the point.
* CORDA-3356 Subflow ledger consistency tests + move statemachine tests to slow integration tests
Add tests for subflows that fail during transitions.
Split out `StatemachineErrorHandlingTest` into a series of smaller tests.
Move these tests into the `integration-test-slow` category so they are
not run against every PR.
* CORDA-3356 Fix detekt issue
* CORDA-3356 Tidy test names
* Added a timestamp property to Checkpoint getting a new Instant.now() value at every Checkpoint instantiation/ copy instantiation. FlowMonitor is now using this new property (Checkpoint#timestamp) and StateMachineState#isFlowResumed to determine which flows are actually suspended. It leaves out flows that are doing work in their FlowLogic#call method.
* Cleaner comment
* Broke FlowMonitor#logFlowsWaitingForParty into logFlowsWaitingForParty and waitingFlowsToDurations. This way waitingFlowsToDurations is modular and can be tested.
Made FlowMonitor constructor get StateMachineManager instead of the retrieveFlows lamda. This way FlowMonitor is more consistent as a service, and entire flow filtering process is now being done in FlowMonitor#waitingFlowsToDurations.
Removed "smm as? StateMachineManagerInternal" in AbstractNode#start as it made no sense.
Updated CheckpointDumper to mention the Checkpoint#timestamp when writing the checkpoint as json.
* Added tests for FlowMonitor service.
* Remove old comment
* 1. FLowMonitor#waitingFlowDurations now returns a Sequence to have an iteration less.
It used to be, one iteration from returning a Set from FLowMonitor#waitingFlowDurations plus one iteration from FlowMonitor#logFlowsWaitingForParty.
2. Code reformattings
* 1. Remove constructor keyword from FlowMonitor
2. Code reformattings
3. Update detekt baseline
* Resolve conflict in Detekt baseline
* Revert "Revert "CORDA-3307 - add support for environment variables in linux (#5523)" (#5643)"
This reverts commit 03ab258fc2.
* Env variables with underscore are now validated using schema validation and checking for unknown key errors.
* Resolving comments from PR review.
* Fix for deprecated import.
* Reworked logic according to PR review.
* Resolved bad string parsing problems where the json structure could be broken if some symbols were included in the key or value.
* Quick and dirty change to stop "Unable to start notaries." error message (#5686)
"Unable to start notaries. A required port might be bound already" is
returned whenever a startup error occurs while starting the notary nodes
in driver tests. This hides the real error.
This change prints the actual error to std_err and read from file
at a later point. This means the real error is not lost and will be
shown in failed builds.
* Suppress detekt warnings
This is to potentially help with debugging in the future as the
`flowId` could become confusing for received messages where the `flowId`
has nothing to do with the current flow.
* * CORDA-2876: Migrate DJVM serialization modules into Corda.
* Pre-generate Corda classes for DJVM sandbox when node boots in production mode.
* Ensure that all DJVM test contract CorDapps are signed.
* Test examining attachments within DJVM sandbox.
* Test Contract.verify() using cryptographic verify function.
* Add test cases for more non-determinism in Contract.verify().
* Update node-driver to support testing nodes with DJVM support.
* Modify Node to allow alternative DJVM configurations for testing.
* Refactor DeterministicVerifierFactoryService for default use-case.
* Small whitespace and code-style refactors.
* Create and activate a DJVM execution profile for the Node.
* Revert making Verifier implement AutoCloseable.
* Allow the node to cache sandboxed Corda byte-code for reuse.
* Use updated Quasar agent that knows not to touch DJVM classloaders.
* Fix Quasar's package exclusions globs for DJVM.
* Deserialise LedgerTransaction into the sandbox for Contract.verify().
* Add the DJVM's serialisation modules to the Corda node.
* Update the node for the latest DJVM API, and preserve the ConstructorForDeserialization annotation on user contract classes.
* Add corda-dev to repositories while DJVM is SNAPSHOT.
* Migrate DJVM specialisation into AbstractNode's ServiceHubInternalImpl.
* Exclude sandbox.** and shaded djvm.** classes from Quasar agent.
* Add the corda-dev repository to :node for the deterministic runtime.
* Turn Verifier into an abstract base class that is specialised by BasicVerifier and DeterministicVerifier.
* Add the Corda deterministic libraries to the Node, and split the DJVM sandbox across two SandboxClassLoader instances.
* Add DJVM to contract verification path inside Corda Node.
* Minor lambda simplifications and removing unused import.
* CORDA-2871: Remove @CordaSerializable from LedgerTransaction.
* CORDA-2871: Add a callback to ServicesForResolution to allow the Node to modify a LedgerTransaction object.
* CORDA-2871: Refactor the contract verification code into a separate class,
and allow LedgerTransaction to choose different Verifier objects.
* Update DJVM to use Corda 4.4-SNAPSHOT. (#95)
* CORDA-3330: Allow DJVM to preload / pregenerate classes from selected jars. (#92)
* Add support for SourceClassLoader.getResources() to DJVM.
* Allow a SandboxConfiguration to preload sandbox byte-code for all classes inside jars containing META-INF/DJVM-preload.
* CORDA-3309: Remove explicit try-catch in favour of UncaughtExceptionHandler. (#91)
* CORDA-3309: Install UncaughtExceptionHandler for DJVM tasks. (#88)
* Fix tests broken by Windows line endings. (#82)
* CORDA-3292: Reimplement ExecutionProfile as a data class. (#80)
* CORDA-2877: Refactor how we create child SandboxConfiguration objects. (#76)
* CORDA-2877: Load bytecode from a persistent cache to prevent repeated rewriting. (#75)
* Refactor byte-code cache to SandboxConfiguration instead of AnalysisConfiguration. We cannot "mix and match" byte-code generated by different sets of rules.
* CORDA-3137: Enhance annotation handling so that we can allow some annotations to be mapped into the sandbox without also needing to be stitched. (#72)
* CORDA-2871: Minor cosmetic fixes. (#69)
* CORDA-3218: Align DJVM with internal Corda Serialisation API. (#68)
* Ensure we get the latest SNAPSHOT of the serialisation code.
* CORDA-2871: Refactor SourceClassLoader to define source classes. (#66)
* Rewrite SourceClassLoader to support parent/child relationships.
* Revert catching TypNotPresebtException - it was a symptom of a bigger problem.
* Remove AutoCloseable from AnalysisConfiguration and SourceClassLoader.
* SourceClassLoader.getResource() must delegate to its parent first.
* CORDA-2871: Ensure ClassLoader.loadClass() throws ClassNotFoundException for all cases where the class cannot be found. (#64)
* CORDA-2871: Modify sandbox tasks to implement both java.Function and sandbox.Function (#62)
* Make TaskExecutors implement BiFunction to make them composable.
* Create ImportTask to wrap a java.Function inside a sandbox.Function.
* Add createExecutor() and createRawExecutor() APIs to SandboxClassLoader.
* Update serialization to use SandboxClassLoader.toSandboxClass().
* Remove a layer of lambdas from the serialisation code.
* Update SandboxExecutor and SandboxRawExecutor.
* Rename Executor to TaskFactory.
* Rename dangling executor -> taskFactory.
* CORDA-2871: Sanity fixes! (#63)
* Improve message for SandboxClassLoadingException.
* Fix serialisation API for using sandboxed environment.
* CORDA-3174: Extend serialisation to include InputStream and OpaqueBytesSubSequence. (#60)
* Update DJVM Example project for serialisation.
* Add serializers for InputStream and OpaqueBytesSubSequence.
* Support ZIP Inflater and CRC32 inside the sandbox.
* Allow the DJVM to wrap java.io.InputStream as sandbox.java.io.InputStream.
* Configure tests also to preserve @DeprecatedConstructorForDeserialization.
* CORDA-3174: Implement Corda serialization modules. (#59)
* Create DJVM serialization modules.
* Create test cases for Array<T>, List<T> and List<Array<T>>.
* Refactor SandboxPrimiveSerializer for all primitive types.
* Implement SandboxCollectionSerializer to support Collection types.
* Implement SandboxMapSerializer to support Map types.
* Attempt to fix infinite loop when computing Collection and Map fingerprints.
* Apply special handling when deserialising sandbox.java.lang.Character.
* Remap Java primitive types to sandbox Java object types to deter evolution.
* Use Class.getPackage().getName() to determine sandbox package name.
* Implement SandboxEnumSerializer to support Enum types.
* Implement SandboxPublicKeySerializer to support Java security keys.
* Add serialization projects to the composite example project.
* Implement serializers for BigInteger, BigDecimal, Currency and StringBuffer.
* Test that deserialising does not instantiate the untrusted user classes.
* Implement serializers for java.time.* types.
* Add serialiser for BitSet - currently disabled until BitSet itself is supported.
* Add serialisers for EnumSet and Class.
* Include support for EnumMap in the SandboxMapSerializer.
* Ensure the DJVM Example project's tests preserve @CordaSerializable.
* Add support for UUID as a primitive type.
* Use common abortReadOnly() method for declaring serialization as unsupported.
* Streamline the API for deserialising into the sandbox.
* Add preliminary support for deserialising X.509 certificates.
* Implement serializer for java.util.Optional.
* Refactor configuration of the sandbox serialization scheme.
* Add tests for deserialising arrays of basic types.
* Include method annotations in annotation stitching. This ensures that `@ConstructorForDeserialization` is not dropped.
* Enable test for SandboxBitSetSerializer.
* Enable tests for X.509 serializers.
* Implement serializers for ProtonJ primitive types.
* Serialize java.util.Date as a primitive type.
* Add the bintray Gradle plugin to the serialisation modules.
* Do not publish serialisation modules - they will become part of Corda itself.
* CORDA-2876: Only apply DJVM sources to Node Driver when devMode=true.
* Resolve DeteKT warnings.
* Require Node's JVM to set -Dnet.corda.djvm=true in order to enable DJVM.
* Enable DJVM for DemoBench nodes.
* Disable Quasar instrumentation verification for DemoBench nodes.
* Upgrade to DJVM 1.0-RC01.
* Try to modify DriverParameters in a more "ABI friendly" way.
* Refactor and simplify sandbox deserialisation of primitive objects.
* Review fixes.
* Update EvolutionSerializerFactory to handle sandboxed primitive boxed types.
* CORDA-3350: Increase size of constraints column (#5639)
* Detekt
* Update api file with new threshold
* Add check in transaction builder
* Revert "Add check in transaction builder"
This reverts commit ca3128f44c.
* Add check for max number of keys
* Update api file
* Address Tudor's comments
* Remove check for pre-5 and add test for EC keys
* fix typo and rename liquibase script
* updated docs with measurement numbers for composite keys
* Make detekt happy again
- Port ledger integrity work to `SingleThreadedStateMachineManager`
- Fix `StatemachineErrorHandlingTest`
- Fix compile errors in `RetryFlowMockTest` + `VaultObserverExceptionTest`
- Add method to `StaffedFlowHospital` that was missed during original merge
* CORDA-3194 Do not allow killed flows back into the hospital
This change has been made to prevent killed flows from being added back
to the hospital after being forcibly removed by `killFlow`. Not doing so,
could leave references to a flow inside of the hospital, which is not
the correct behaviour.
`killFlow` now sets a flow's `StatemachineState.isRemoved` to true.
This check is then used in `StaffedFlowHospital` and the
`DumpHistoryOnErrorInterceptor`.
* CORDA-3194 Log different message for transition error due to killed flow
When a flow is killed, its checkpoint is deleted. Currently, the
statemachine will still try a process the next event even if it has
been killed. This can lead to an error when trying to update the
deleted checkpoint. The exception thrown from this is logged out.
An if statement has been added to log a different message at debug level
if it is due to an update error for a killed flow. This is done to not
alarm node operators of the exception.
* CORDA-3194 Relax duplicate insert flow hospital handling
Revert a previous change to now make the duplicate insert staff member to
give a diagnosis of discharge or not my speciality (previously gave
terminal).
This is to prevent duplicate insert handling from overriding finality
flow error handling.
* added the warning as a TimerTask at StaffedFlowHospital#delayedDischargeTimer
* moved the scheduling of the warning task at StaffedFlowHospital#init block. That way we ensure that the task will be scheduled only once at StaffedFlowHospital initialization.
* Corrected overnight observation warning task's logging message. Changed StaffedFlowHospital#delayedDischargeTimer to the more generic StaffedFlowHospital#hospitalJobTimer since it now schedules delayed discharges tasks as well the overnight observation warning task. Removed this from property reference
* switching to fun timerTask for the instantiation of anonymous TimerTask classes
* Correct condition to log patients who are currently in the hospital, whose last record in their medical records is Outcome.OVERNIGHT_OBSERVATION. Extended logging to include treatableSessionInits staying in the hospital
* Add not empty check for patientsUnderOvernightObservation. Correct strings.
When a flow fails to retry, it should be kept in for overnight observation and aborted.
In the future, it might be possible to retry flows again that failed during their retry, but for now keeping for observation and aborting is satisfactory.
* CORDA-3194 Remove hospitalised flows from `HospitalisingInterceptor`
Small refactor to remove some of the hospital logic out of the
`HospitalisingInterceptor` and into the `StaffedFlowHospital`.
Add some comments to help clarify the purpose of the two maps inside
of the hospital.
* CORDA-3194 When a flow fails to retry force it into observation
When a flow fails to retry, it should be kept in for overnight
observation and aborted.
In the future, it might be possible to retry flows again that failed
during their retry, but for now keeping for observation and aborting is
satisfactory.
* CORDA-3194 Test for database commit failure when retrying a flow
Failing during the database commit failure that occurs after the retry
flow action does not stop the flow from actually retrying. This test
confirms this functionality.
The retried flow gets scheduled as part of the retry action. The failure
in the commit action does not prevent this since it has already been
scheduled.
* CORDA-3194 Replay start flow events when responding flow fails initial checkpoint commit
Logic has already been added to recover from initial checkpoint commit
failures on the initiating flow side but this did not suffice for
the same failure occurring on the responding flow's side.
The same idea has been added to resolve the responding flow's issue.
`ExternalMessageEvent` now has a `flowId` that is maintained on the
event. Messages can then be replayed to start/restart the flow, while
the event provides the flow id to each flow start.
Each `ExternalMessageEvent` implementation generates a random `flowId`
when constructed.
Events are stored in Artemis. This allows the solution to recover across
node restarts as the events will be pulled from artemis again when
restarting.
In the future `flowId`s will probably moved off of the events and
generated purely on the responding flow's node.
* CORDA-3194 Add test to verify that errors removing a responding flow are recoverable
* Unwrap rx.OnErrorNotImplementedException so the hospital can handle the cause appropriately
* Add db failure cordapp
* Renamed folders to avoid ambiguity in gradle
* Add integration test for exception hospitalisation when thrown from an RX observable.
* Make the test slightly cleaner
* Fix the schema to actually match the requirements for my custom state. Thanks a bunch, H2.
* Switch test to use SqlException base class.
* Schedule error event if we detect that a commit or db flush has thrown (forcing the flow to error even if customer code then goes ahead to swallow the exception)
* Revert change to schedule extra error
* Add more tests for edge case with DB exceptions, changed CorDapp to suppor this an hook in the flow hospital
* Warning about unsubscribe
Check state transitioned from clean to error for hospital admission.
* Match the test to our actual expectations
* Revert "Revert change to schedule extra error"
This reverts commit 43d47937
* Prevent suppression of errors arising in `transaction()` and `jdbcConnection()`
* Test for SqlException caught trying to escape from recordTransaction and suppressed outside being intercepted.
* More tests for various error/catch combinations
* Clean up and comments
* Code reformat
* Fix test compilation
Wrap exceptions that occur in state machine transitions with a custom exception type which is
then handled inside of the flow hospital. As part of this change, a number of side negative side
effects have been addressed.
General summary:
- `StateTransitionException` wraps exceptions caught in `TransitionExecutorImpl`
- `StateTransitionExceptions` are handled in the flow hospital, retried 3 times and then kept in
for observation if errors persist (assuming conditions below are false)
- Exceptions that occur in `FlowAsyncOperation` events are wrapped in
`AsyncOperationTransitionException` and ignored by the flow hospital transition staff member
- `InterruptException`s are given a `TERMINAL` diagnosis by the flow hospital transition staff
member (can occur due to `killFlow`)
- Allow flows which have not persisted their original checkpoint to still retry by replaying their
start flow messages
- Swallow exceptions in `AcknowledgeMessages` actions
Detailed summary:
* CORDA-3194 Add state machine transition error handling to flow hospital
Wrap exceptions that are caught in `TransitionExecutorImpl` (coming from
new errors) with `StateTransitionException`. This exception is then
handled explicitly by the flow hospital.
Add `TransitionErrorGeneralPractitioner` to `StaffedFlowHospital`. This
staff member handles errors that mention `StateTransitionException`.
Errors are retried and then kept in the hospital if the errors persist.
* CORDA-3194 Remove a fiber from the `hospitalisedFlows` if its previous state was clean
If the fiber's previous state was clean then remove it from
`HospitalisingInterceptor.hospitalisedFlows`. This allows flows that are
being retried to clean themselves. Doing this allows them to re-enter
the flow hospital after executing the fiber's transition (if an error
occurs).
This is important for retrying a flow that has errored during a
transition.
* CORDA-3194 Set `isAnyCheckpointPersisted` to true when retrying a flow
Added to prevent a single flow from creating multiple checkpoints when
a failure occurs during `Action.AcknowledgeMessages`.
More specifically, to `isAnyCheckpointPersisted` is false when retrying
the flow, even though a checkpoint has actually been saved. Due to this
a brand new flow is started with a new flow id (causing duplication).
Setting `isAnyCheckpointPersisted` to true specifically when retrying a
flow resolves this issue.
* CORDA-3194 Add Byteman test to verify transition error handling
Add `StatemachineErrorHandlingTest` to verify transition error handling.
Byteman allows exceptions to be injected at certain points in the code's
execution. Therefore exceptions can be thrown when needed inside of the
state machine.
The current tests check errors in events:
- `InitiateFlow`
- `AcknowledgeMessages`
* CORDA-3194 Swallow all exceptions in `ActionExecutorImpl.executeAcknowledgeMessages`
Swallow the exceptions that occur in the `DeduplicationHandler`s when
inside of `ActionExecutorImpl.executeAcknowledgeMessages`.
The side effects of the failures that can happen in the handlers are
not serious enough to put the transition into a failure state.
Therefore they are now caught. This allows the transition to continue
as normal, even if an error occurs in one any of the handlers.
* CORDA-3194 Wrap unexpected exceptions thrown in async operation transitions
Exceptions thrown inside of `FlowAsyncOperation.execute` implementations
that are not returned as part of the future, are caught, wrapped and
rethrown. This prevents unexpected exceptions thrown by (most likely)
user code from being handled by the hospital by the transition
staff member.
This handling might change moving forward, but it allows the async
operation to continue working as it was before transition error handling
was added.
* CORDA-3194 Verify that errors inside of `AcknowledgeMessages` work as expected
Update `StatemachineErrorHandlingTest` to correctly test errors that
occur when executing the `AcknowledgeMessages` action.
* CORDA-3194 Retry flows that failed to persist their original checkpoint
Allow a flow that failed when creating their original checkpoint (for
example - failing to commit the db transaction) to retry.
The flow will create a brand new checkpoint (as the original did not
saved).
This required adding `flowId` to `ExternalStartFlowEvent` to allow the
event to keep a record of the flow's id. When the flow is retried, the
events are replayed which trigger a flow to be started that has the
id stored in the event.
To allow this change, code was removed from `retryFlowFromSafePoint` to
allow the function to continue, even if no checkpoint matches the passed
in flow id.
* CORDA-3194 Correct `FlowFrameworkTests` test due to error handling
Test assumed that errors in transitions are not retried, this has now
been updated so the test passes with the flow succeeding after an
exception is thrown.
* CORDA-3194 Remove unneeded import
* CORDA-3194 Make the state transition exceptions extend `CordaException`
`StateTransitionException` and `AsyncOperationTransitionException` now
extend `CordaException` instead of `Exception`.
* CORDA-3194 Improve log messages
* CORDA-3194 Remove unneeded code in `HospitalisingInterceptor`
Due to a previous change, a section of code that removes a flow id
from the `hospitalisedFlows` map is no longer required. This code has
been removed.
* CORDA-3194 Constraint violations are given `TERMINAL` diagnosis
Add `Diagnosis.TERMINAL` to `StaffedFlowHospital` to allow an error
to be ignored and left to die a quick and painful death.
`StateTransitionException` changed so it does not cause serialisation
errors when propagated from a flow.
* CORDA-3194 `InterruptedExceptions` are given `TERMINAL` diagnosis
* Add GP to flow hospital, and start working on a list of things the GP knows to be incurable.
* Only hospitalise SQL and Persistence Exceptions (let's see if that is enough?), also rename to DatabaseDentist.
* Disabled hospitalisation of SQL exceptions in flow retry tests
* Fix RPC exception handling test by not using PersistenceException
* Ignore flaky integration test
* Code review: Rename staff member and add testing annotation
* Revert compiler.xml
