chore: initial import from KNEL/PFVCluster@041d311 [#769]

Split per O&M lane work order. Full history: KNEL/PFVCluster.
https://projects.knownelement.com/issues/769#note-4152
This commit is contained in:
2026-09-03 21:39:22 -05:00
commit 51a1c9deee
41 changed files with 3428 additions and 0 deletions
+20
View File
@@ -0,0 +1,20 @@
#!/bin/bash
# garage-pdu-relay: expose APC AP7830 metered output current via snmpd extend.
#
# Relay pattern (founder ruling 2026-09-02): the garage PDU speaks SNMPv1
# only, walks die on its old AOS, and its ACL admits the LibreNMS VM vantage.
# This relay runs ON tsys-librenms; HA polls the relay via the fleet snmpd
# extend (same rails as drac-extend). Output: "output_amps <x.y>".
set -euo pipefail
PDU_HOST="${PDU_HOST:-pfv-garage-pdu-1.knel.net}"
COMMUNITY="${PDU_COMMUNITY:-kn3lmgmt}"
OID=".1.3.6.1.4.1.318.1.1.12.2.3.1.1.2.1"
raw=$(snmpget -v1 -c "$COMMUNITY" -On -Ov "$PDU_HOST" "$OID" 2>/dev/null \
| grep -oE '[0-9]+$') || true
if [ -z "${raw:-}" ]; then
echo "output_amps unavailable"
exit 0
fi
awk -v t="$raw" 'BEGIN { printf "output_amps %.1f\n", t/10 }'
+146
View File
@@ -0,0 +1,146 @@
#!/bin/bash
# Deploy lm-sensors extend (native sensors + TEMPer) + snmpd on a host [#341/#439]
#
# Run ON the target host via the remote.sh chokepoint (bash -s). Copy the
# wrapper over first (vm-copy / prox-copy). Idempotent: safe to re-run.
#
# Native lm-sensors (CPU/board temps, fans) is installed on every target.
# The TEMPer USB layer (venv + temperusb + udev rule) is installed ONLY when
# a 0c45:7401 probe is attached.
#
# Usage (from repo root):
# VM_IP=stlpc-artroom VM_USER=root bash tests/remote.sh vm-copy dcinfra/sensors/temper/lmsensors-extend.sh /tmp/lmsensors-extend.sh
# VM_IP=stlpc-artroom VM_USER=root bash tests/remote.sh vm-file dcinfra/sensors/temper/deploy.sh
# PROX_HOST=pfv-tsys1 bash tests/remote.sh prox-copy dcinfra/sensors/temper/lmsensors-extend.sh /tmp/lmsensors-extend.sh
# PROX_HOST=pfv-tsys1 bash tests/remote.sh prox-file dcinfra/sensors/temper/deploy.sh
#
# snmpd is a targeted exception: sensor hosts + tsys-* Proxmox hosts
# (founder-directed 2026-08-27), not fleet-wide.
set -euo pipefail
COMMUNITY="${SNMP_COMMUNITY:-kn3lmgmt}"
VENV_DIR="${VENV_DIR:-/opt/temper-venv}"
WRAP_SRC="${WRAP_SRC:-/tmp/lmsensors-extend.sh}"
WRAP_DST="/usr/local/bin/lmsensors-extend"
WRAP_DST_OLD="/usr/local/bin/temper-lmsensors"
SNMPD_CONF="/etc/snmp/snmpd.conf"
UDEV_RULE="/etc/udev/rules.d/99-temper.rules"
MARK_BEGIN="# lmsensors-extend managed block begin"
MARK_END="# lmsensors-extend managed block end"
OLD_MARK_BEGIN="# temper-lmsensors managed block begin"
OLD_MARK_END="# temper-lmsensors managed block end"
# Source allowlist (founder ruling 2026-08-27: bind Tailscale-only where
# possible; HA is a Tailscale APP connector so its polls source from its LAN
# IP and must be ACL'd explicitly — no broad subnet grants).
# 192.168.3.176 tsys-librenms LAN (poller)
# 100.86.204.77 tsys-librenms Tailscale
# 192.168.3.12 pfv-bms / Home Assistant LAN (app connector source)
# 100.67.108.125 pfv-bms Tailscale
# 100.101.187.119 admin workstation Tailscale
# 152.53.37.179 + 2a0a:4cc0:2000:686::a496:3bff:fe35:610a Cloudron VPS
# (future Uptime Kuma SNMP polling)
SNMP_SOURCES="${SNMP_SOURCES:-192.168.3.176 192.168.3.252 192.168.3.253 100.70.181.72 100.71.171.20 100.86.204.77 192.168.3.12 100.67.108.125 100.101.187.119 152.53.37.179 2a0a:4cc0:2000:686:a496:3bff:fe35:610a}"
log() { echo "[deploy] $*"; }
has_probe() {
command -v lsusb >/dev/null 2>&1 && lsusb | grep -qi '0c45:7401'
}
host_lan_ip() {
# first 192.168.* address (mgmt/user LAN; excludes 127/10.100 storage/dockers)
ip -4 addr show 2>/dev/null | grep -oP 'inet \K192\.168\.[0-9.]+' | head -1
}
host_ts_ip() {
command -v tailscale >/dev/null 2>&1 && tailscale ip -4 2>/dev/null | head -1
}
[[ $EUID -eq 0 ]] || { echo "must run as root" >&2; exit 1; }
[[ -f "$WRAP_SRC" ]] || { echo "wrapper not found at $WRAP_SRC (copy it first)" >&2; exit 1; }
# --- native lm-sensors layer (all hosts) ---
if ! dpkg -s lm-sensors >/dev/null 2>&1; then
log "installing lm-sensors"
DEBIAN_FRONTEND=noninteractive apt-get update -qq
DEBIAN_FRONTEND=noninteractive apt-get install -y -qq lm-sensors
fi
if [[ -z "$(sensors 2>/dev/null)" ]]; then
log "no sensor readings yet; trying coretemp/k10temp modules"
modprobe coretemp 2>/dev/null || true
modprobe k10temp 2>/dev/null || true
fi
log "native sensors: $(sensors 2>/dev/null | grep -c '°C') temp readings"
# --- TEMPer USB layer (probe-conditional) ---
if has_probe; then
log "TEMPer probe present: installing venv layer"
if ! dpkg -s python3-venv >/dev/null 2>&1; then
log "installing python3-venv"
DEBIAN_FRONTEND=noninteractive apt-get update -qq
DEBIAN_FRONTEND=noninteractive apt-get install -y -qq python3-venv
fi
if [[ ! -x "$VENV_DIR/bin/pip" ]]; then
[[ -d "$VENV_DIR" ]] && { log "removing broken venv $VENV_DIR"; rm -rf "$VENV_DIR"; }
log "creating venv $VENV_DIR"
python3 -m venv "$VENV_DIR"
fi
log "ensuring temperusb in venv"
"$VENV_DIR/bin/pip" install --quiet --disable-pip-version-check temperusb
# TEMPer USB dongle must be readable by the unprivileged snmpd user
# (extend runs as Debian-snmp; pyusb needs the raw usb device node).
log "installing udev rule for 0c45:7401"
cat >"$UDEV_RULE" <<'EOF'
SUBSYSTEM=="usb", ATTRS{idVendor}=="0c45", ATTRS{idProduct}=="7401", MODE="0666"
EOF
udevadm control --reload-rules
udevadm trigger
else
log "no TEMPer probe: skipping venv/udev layer"
fi
log "installing wrapper to $WRAP_DST"
install -m 0755 "$WRAP_SRC" "$WRAP_DST"
rm -f "$WRAP_DST_OLD"
if ! dpkg -s snmpd >/dev/null 2>&1 || ! command -v snmpget >/dev/null 2>&1; then
log "installing snmpd + snmp client"
DEBIAN_FRONTEND=noninteractive apt-get update -qq
DEBIAN_FRONTEND=noninteractive apt-get install -y -qq snmpd snmp
fi
log "configuring snmpd (community=$COMMUNITY, scoped sources)"
LAN_IP="$(host_lan_ip)"
TS_IP="$(host_ts_ip)"
AGENT_ADDRS=""
[[ -n "$LAN_IP" ]] && AGENT_ADDRS="udp:${LAN_IP}:161"
[[ -n "$TS_IP" ]] && AGENT_ADDRS="${AGENT_ADDRS:+$AGENT_ADDRS,}udp:${TS_IP}:161"
[[ -n "$AGENT_ADDRS" ]] || { echo "no LAN/Tailscale address found; refusing to bind 0.0.0.0" >&2; exit 1; }
sed -i "/^$OLD_MARK_BEGIN\$/,/^$OLD_MARK_END\$/d" "$SNMPD_CONF"
sed -i "/^$MARK_BEGIN\$/,/^$MARK_END\$/d" "$SNMPD_CONF"
sed -i 's/^agentaddress/#agentaddress/' "$SNMPD_CONF"
{
echo "$MARK_BEGIN"
echo "agentaddress $AGENT_ADDRS"
for src in $SNMP_SOURCES; do
if [[ "$src" == *:* ]]; then echo "rocommunity6 $COMMUNITY $src"; else echo "rocommunity $COMMUNITY $src"; fi
done
echo "extend lmsensors $WRAP_DST"
echo "extend lmsensors_n $WRAP_DST -n"
echo "$MARK_END"
} >>"$SNMPD_CONF"
log "agentaddress: $AGENT_ADDRS"
log "restarting snmpd"
systemctl enable --now snmpd >/dev/null
systemctl restart snmpd
log "wrapper output:"
"$WRAP_DST"
log "snmp extend self-test (expect lm-sensors block below):"
sleep 1
snmpget -v2c -c "$COMMUNITY" -Ovq localhost \
.1.3.6.1.4.1.8072.1.3.2.3.1.2.9.108.109.115.101.110.115.111.114.115 2>/dev/null || \
log "WARN: snmpget self-test failed (check snmpd logs)"
log "done"
+196
View File
@@ -0,0 +1,196 @@
#!/bin/bash
# lm-sensors extend for snmpd: native sensors + TEMPer USB block [#341/#439]
#
# Deployed to /usr/local/bin/lmsensors-extend on sensor hosts by deploy.sh.
# Output = lm-sensors format, consumed by Home Assistant's SNMP integration
# (nsExtendOutputFull."lmsensors") and, if enabled later, LibreNMS's
# lm-sensors app.
#
# Layers (each optional, at least one required):
# 1. native /usr/bin/sensors output (CPU/board temps, fans) — lm-sensors pkg
# 2. TEMPer USB probe (temperusb in venv /opt/temper-venv) — probe-optional
# (venv-on-host is founder-approved for sensor hosts, incl. Proxmox)
set -euo pipefail
VENV_PY="${VENV_PY:-/opt/temper-venv/bin/python}"
NATIVE_SENSORS_BIN="${NATIVE_SENSORS_BIN:-/usr/bin/sensors}"
SYSFS_THERMAL_ROOT="${SYSFS_THERMAL_ROOT:-/sys/class/thermal}"
CHIP_NAME="${CHIP_NAME:-temper-usb-1}"
# Render lm-sensors format block from internal/external Celsius values.
# Empty/absent external renders internal only. Non-numeric input is rejected.
format_block() {
local internal="${1:-}" external="${2:-}"
local num_re='^-?[0-9]+(\.[0-9]+)?$'
[[ "$internal" =~ $num_re ]] || {
echo "error: internal temp not numeric: '$internal'" >&2
return 1
}
local signed
printf '%s\nAdapter: USB adapter\n' "$CHIP_NAME"
signed="${internal#+}"; [[ "$signed" == -* ]] || signed="+$signed"
printf 'Internal: %s°C\n' "$signed"
if [[ -n "$external" && "$external" =~ $num_re ]]; then
signed="${external#+}"; [[ "$signed" == -* ]] || signed="+$signed"
printf 'External: %s°C\n' "$signed"
fi
}
# True when a TEMPer USB dongle (0c45:7401) is attached and the venv exists.
# Numeric mode: print ONE Celsius float for snmpd extend lmsensors_n [#618].
# Preference: TEMPer external > TEMPer internal > first native/sysfs temp
# reading. Consumed by HA's SNMP config-entry integration (device_class
# temperature; HA renders C in the instance display unit).
numeric_pick() {
local internal="${1:-}" external="${2:-}" native="${3:-}" pick
local num_re='^-?[0-9]+(\.[0-9]+)?$'
for pick in "$external" "$internal"; do
pick="${pick#+}"
[[ "$pick" =~ $num_re ]] && { printf '%s\n' "$pick"; return 0; }
done
if [[ -n "$native" ]]; then
pick="$(grep -oE "\-?[0-9]+(\.[0-9]+)?°C" <<<"$native" | head -1)"; pick="${pick%°C}"
[[ "$pick" =~ $num_re ]] && { printf '%s\n' "$pick"; return 0; }
fi
return 1
}
numeric_main() {
local internal external line
if has_temper_probe; then
if read -r internal external <<<"$(read_temper 2>/dev/null)"; then
numeric_pick "${internal:-}" "${external:-}" "" && return 0
fi
fi
line="$(run_native_sensors | grep -m1 '°C')"
[[ -n "$line" ]] && numeric_pick "" "" "$line" && return 0
line="$(read_sysfs_thermal 2>/dev/null | head -1 | awk '{printf "+%.1f°C", $2/1000}')"
[[ -n "$line" ]] && numeric_pick "" "" "$line" && return 0
echo "error: numeric mode found no sensor source" >&2
return 1
}
has_temper_probe() {
command -v lsusb >/dev/null 2>&1 || return 1
[[ -x "$VENV_PY" ]] || return 1
lsusb | grep -qi '0c45:7401'
}
# Read the TEMPer probe via temperusb; prints "<internal> <external>" in C.
# Sensor 0 = internal, sensor 1 = external probe (if attached).
read_temper() {
"$VENV_PY" - <<'PY'
from temperusb import TemperHandler
internal = external = ""
for device in TemperHandler().get_devices():
try:
temps = device.get_temperatures(sensors=[0, 1])
except (ValueError, KeyError):
temps = device.get_temperatures()
for index, slot in ((0, "internal"), (1, "external")):
reading = temps.get(index)
if reading and reading.get("temperature_c") is not None:
value = f"{reading['temperature_c']:.1f}"
if slot == "internal":
internal = value
else:
external = value
print(f"{internal} {external}".strip())
PY
}
run_native_sensors() {
[[ -x "$NATIVE_SENSORS_BIN" ]] || return 1
"$NATIVE_SENSORS_BIN" 2>/dev/null
}
# SBC fallback (Raspberry Pi et al.): lm-sensors often reports nothing; SoC
# temperatures live in /sys/class/thermal/thermal_zone*/{type,temp} (mC).
read_sysfs_thermal() {
[[ -d "$SYSFS_THERMAL_ROOT" ]] || return 1
local z type mc found=0
for z in "$SYSFS_THERMAL_ROOT"/thermal_zone[0-9]*; do
[[ -r "$z/type" && -r "$z/temp" ]] || continue
type="$(<"$z/type")" mc="$(<"$z/temp")"
[[ "$mc" =~ ^-?[0-9]+$ ]] || continue
printf '%s %s\n' "$type" "$mc"
found=1
done
((found)) || return 1
}
# Render lm-sensors format block from read_sysfs_thermal lines "type millicelsius".
format_sysfs_block() {
local input="${1:-}" type mc signed n=0
[[ -n "$input" ]] || return 1
printf 'soc-thermal-virtual-0\nAdapter: Virtual device\n'
while read -r type mc; do
((++n))
signed="$(awk -v mc="$mc" 'BEGIN { printf "%.1f", mc / 1000 }')"
signed="${signed#+}"
[[ "$signed" == -* ]] || signed="+$signed"
printf 'temp%d: %s°C\n' "$n" "$signed"
done <<<"$input"
}
# Labeled per-zone temps (e.g. Jetson GPU-therm) [#736]. Hosts opt in by
# listing zone types (one per line) in /etc/snmp/lmsensors-zone-types; no
# snmpd.conf change or restart needed (extend re-execs per poll). Each type
# renders its own chip block so HA templates can anchor on the chip name.
ZONE_TYPES_FILE="${ZONE_TYPES_FILE:-/etc/snmp/lmsensors-zone-types}"
format_labeled_zones() {
local z type mc want signed
[[ -r "$ZONE_TYPES_FILE" ]] || return 1
while IFS= read -r want; do
want="${want%%#*}"; want="${want//[[:space:]]/}"
[[ -n "$want" ]] || continue
for z in "$SYSFS_THERMAL_ROOT"/thermal_zone[0-9]*; do
[[ -r "$z/type" && -r "$z/temp" ]] || continue
[[ "$(<"$z/type")" == "$want" ]] || continue
mc="$(<"$z/temp")"
[[ "$mc" =~ ^-?[0-9]+$ ]] || continue
signed="$(awk -v mc="$mc" 'BEGIN { printf "%.1f", mc / 1000 }')"
signed="${signed#+}"
[[ "$signed" == -* ]] || signed="+$signed"
printf 'zone-%s-virtual-0\nAdapter: Virtual device\ntemp1: %s°C\n\n' "$want" "$signed"
done
done <"$ZONE_TYPES_FILE"
return 0
}
main() {
local out="" block internal external native="" sysinput
if native="$(run_native_sensors)" && [[ -n "$native" ]]; then
out+="$native"$'\n'
else
native=""
fi
if has_temper_probe; then
if block="$(read_temper 2>/dev/null)" \
&& read -r internal external <<<"$block" \
&& block="$(format_block "${internal:-}" "${external:-}")"; then
out+="$block"$'\n'
else
echo "warn: TEMPer probe present but read failed" >&2
fi
fi
if [[ -z "$native" ]] \
&& sysinput="$(read_sysfs_thermal 2>/dev/null)" \
&& block="$(format_sysfs_block "$sysinput")"; then
out+="$block"$'\n'
fi
if block="$(format_labeled_zones)" && [[ -n "$block" ]]; then
out+="$block"$'\n'
fi
[[ -n "$out" ]] || {
echo "error: no sensor sources available (lm-sensors pkg? TEMPer venv? sysfs thermal?)" >&2
return 1
}
printf '%s' "$out"
}
if [[ "${BASH_SOURCE[0]}" == "$0" ]]; then
if [[ "${1:-}" == "-n" ]]; then numeric_main; else main; fi
fi