mirror of
https://github.com/casjaysdevdocker/forgejo
synced 2026-10-15 14:12:55 -04:00
fix: harden Forgejo runner lifecycle
- README.md: document full runner configuration, isolated state, and cleanup settings - IDEA.md: N/A - no user-facing product behavior change - TODO.AI.md: record runner cleanup and runtime verification status - rootfs/usr/local/bin/cleanup-runners: add safe stale Docker and act-cache cleanup - rootfs/usr/local/bin/start-runners: propagate per-runner config and report daemon failures - rootfs/usr/local/etc/docker/init.d/05-dockerd.sh: seed Docker config without overwriting user files - rootfs/usr/local/etc/docker/init.d/08-forgejo.sh: seed Forgejo config without overwriting user files - rootfs/usr/local/etc/docker/init.d/zz-act_runner.sh: seed runner config safely and schedule validated cleanup
This commit is contained in:
7 files changed
+227
-23
No files matched your search
Executable
+95
@@ -0,0 +1,95 @@
|
||||
#!/usr/bin/env bash
|
||||
# shellcheck shell=bash
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
##@Version : 202610051200-git
|
||||
# @@Author : Jason Hempstead
|
||||
# @@Contact : jason@casjaysdev.pro
|
||||
# @@License : LICENSE.md
|
||||
# @@ReadME : cleanup-runners --help
|
||||
# @@Copyright : Copyright: (c) 2026 Jason Hempstead, Casjays Developments
|
||||
# @@Description : Prune docker and act cache leftovers from act runner jobs
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
VERSION="202610051200-git"
|
||||
set -eo pipefail
|
||||
|
||||
RUNNER_CLEANUP_UNTIL="${RUNNER_CLEANUP_UNTIL:-72h}"
|
||||
RUNNER_CLEANUP_DISK_PERCENT="${RUNNER_CLEANUP_DISK_PERCENT:-80}"
|
||||
RUNNER_CLEANUP_DOCKER_DIR="${RUNNER_CLEANUP_DOCKER_DIR:-/data/docker}"
|
||||
RUNNER_CLEANUP_ACT_CACHE_DAYS="${RUNNER_CLEANUP_ACT_CACHE_DAYS:-7}"
|
||||
RUNNER_CLEANUP_ACT_CACHE_DIRS="${RUNNER_CLEANUP_ACT_CACHE_DIRS:-$HOME/.cache/act /root/.cache/act}"
|
||||
RUNNER_CLEANUP_LOG_FILE="${RUNNER_CLEANUP_LOG_FILE:-/data/logs/act_runner/cleanup.log}"
|
||||
if ! [[ "$RUNNER_CLEANUP_DISK_PERCENT" =~ ^(100|[0-9]{1,2})$ ]]; then
|
||||
RUNNER_CLEANUP_DISK_PERCENT=80
|
||||
fi
|
||||
|
||||
__log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*" >>"$RUNNER_CLEANUP_LOG_FILE" 2>&1; }
|
||||
__prune() {
|
||||
local output=""
|
||||
if output=$("$@" 2>&1); then
|
||||
__log "$* :: $(printf '%s\n' "$output" | grep -i -- 'reclaimed' | tail -n1)"
|
||||
else
|
||||
__log "FAILED: $* :: $output"
|
||||
fi
|
||||
}
|
||||
__disk_percent() {
|
||||
df -P "$1" 2>/dev/null | awk 'NR==2 {gsub("%", "", $5); print $5}'
|
||||
}
|
||||
__help() {
|
||||
cat <<'EOF'
|
||||
Usage: cleanup-runners [--help|--version]
|
||||
|
||||
Prune stale Docker runner resources, build cache, images, and act runner
|
||||
cache entries. Only explicitly dangling volumes matching the act- prefix are removed.
|
||||
|
||||
Environment:
|
||||
RUNNER_CLEANUP_UNTIL Docker prune age (default: 72h)
|
||||
RUNNER_CLEANUP_DISK_PERCENT Image cleanup threshold (default: 80)
|
||||
RUNNER_CLEANUP_DOCKER_DIR Docker data path (default: /data/docker)
|
||||
RUNNER_CLEANUP_ACT_CACHE_DAYS Cache age in days (default: 7)
|
||||
RUNNER_CLEANUP_ACT_CACHE_DIRS Space-separated act cache directories
|
||||
RUNNER_CLEANUP_LOG_FILE Cleanup log path
|
||||
EOF
|
||||
}
|
||||
|
||||
case "${1:-}" in
|
||||
-h|--help)
|
||||
__help
|
||||
exit 0
|
||||
;;
|
||||
--version)
|
||||
echo "$VERSION"
|
||||
exit 0
|
||||
;;
|
||||
esac
|
||||
|
||||
mkdir -p "$(dirname -- "$RUNNER_CLEANUP_LOG_FILE")"
|
||||
if ! docker info >/dev/null 2>&1; then
|
||||
__log "docker is not reachable, skipping Docker cleanup"
|
||||
else
|
||||
__log "Starting cleanup"
|
||||
__prune docker container prune -f --filter "until=$RUNNER_CLEANUP_UNTIL"
|
||||
__prune docker network prune -f --filter "until=$RUNNER_CLEANUP_UNTIL"
|
||||
act_volumes="$(docker volume ls -q --filter dangling=true --filter name=act- 2>/dev/null | grep -E -- '^act-' || true)"
|
||||
if [ -n "$act_volumes" ]; then
|
||||
# shellcheck disable=SC2086
|
||||
__prune docker volume rm $act_volumes
|
||||
fi
|
||||
__prune docker builder prune -af --filter "until=$RUNNER_CLEANUP_UNTIL"
|
||||
__prune docker image prune -f --filter "until=$RUNNER_CLEANUP_UNTIL"
|
||||
disk_used="$(__disk_percent "$RUNNER_CLEANUP_DOCKER_DIR" || true)"
|
||||
if [[ "$disk_used" =~ ^[0-9]+$ ]] && [ "$disk_used" -ge "$RUNNER_CLEANUP_DISK_PERCENT" ]; then
|
||||
__log "$RUNNER_CLEANUP_DOCKER_DIR is ${disk_used}% used (limit ${RUNNER_CLEANUP_DISK_PERCENT}%), removing unused images"
|
||||
__prune docker image prune -af --filter "until=$RUNNER_CLEANUP_UNTIL"
|
||||
fi
|
||||
fi
|
||||
|
||||
if [[ "$RUNNER_CLEANUP_ACT_CACHE_DAYS" =~ ^[1-9][0-9]*$ ]]; then
|
||||
for act_cache_dir in $RUNNER_CLEANUP_ACT_CACHE_DIRS; do
|
||||
[ -d "$act_cache_dir" ] || continue
|
||||
removed="$(find "$act_cache_dir" -mindepth 1 -maxdepth 1 -mtime +"$RUNNER_CLEANUP_ACT_CACHE_DAYS" 2>/dev/null | wc -l || true)"
|
||||
find "$act_cache_dir" -mindepth 1 -maxdepth 1 -mtime +"$RUNNER_CLEANUP_ACT_CACHE_DAYS" -exec rm -rf {} + 2>/dev/null || true
|
||||
__log "Removed $removed stale entries from $act_cache_dir"
|
||||
done
|
||||
fi
|
||||
__log "Cleanup finished"
|
||||
exit 0
|
||||
@@ -28,6 +28,34 @@ echo "${emoji}Fatal error occurred: Exit code $retVal at line $LINENO in command
|
||||
kill -TERM 1' ERR
|
||||
trap 'retVal=$?;if [ "$SERVICE_IS_RUNNING" != "yes" ] && [ -f "$SERVICE_PID_FILE" ]; then rm -Rf "$SERVICE_PID_FILE"; fi;exit $retVal' SIGINT SIGTERM SIGPWR
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
__help() {
|
||||
cat <<'EOF'
|
||||
Usage: start-runners
|
||||
|
||||
Register and start one or more Forgejo act_runner daemons in parallel.
|
||||
|
||||
Environment:
|
||||
SERVER_ADDRESS Forgejo server URL (defaults to http://$HOSTNAME)
|
||||
SERVER_TOKEN Required runner registration token
|
||||
RUNNERS_LOG_DIR Runner log directory
|
||||
RUNNERS_START Number of runners to start (default: 1)
|
||||
RUNNER_NAME_PREFIX Runner name prefix (default: runner)
|
||||
RUNNER_LABELS Comma-separated runner labels
|
||||
RUNNERS_CONFIG_BASE Base act_runner config file
|
||||
RUNNER_CACHE_HOST External cache host (used with RUNNER_CACHE_PORT and
|
||||
RUNNER_CACHE_SECRET when no base config is available)
|
||||
RUNNER_CACHE_PORT External cache port
|
||||
RUNNER_CACHE_SECRET External cache secret
|
||||
EOF
|
||||
}
|
||||
|
||||
case "${1:-}" in
|
||||
-h|--help)
|
||||
__help
|
||||
exit 0
|
||||
;;
|
||||
esac
|
||||
|
||||
# Function to __log messages with timestamp
|
||||
__log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $1" >>"$RUNNERS_LOG_DIR/runners" 2>&1; }
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
@@ -100,29 +128,55 @@ __log "Starting $RUNNERS_START act_runner instance(s)"
|
||||
__log "Server Address: $SERVER_ADDRESS"
|
||||
__log "Runner Name Prefix: ${RUNNER_NAME_PREFIX:-runner}"
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
# Shared cache config so runner-N instances use the same cache-server
|
||||
# started by zz-act_runner.sh instead of an unshared per-process cache.
|
||||
# Full base config rendered by zz-act_runner.sh. Fall back to a shared cache-only
|
||||
# config when this script is invoked without a rendered Forgejo runner config.
|
||||
RUNNERS_DEFAULT_HOME="${RUNNER_DEFAULT_HOME:-/config/act_runner/forgejo}"
|
||||
# Keep the established registration path by default; callers may override it
|
||||
# independently when relocating runner state.
|
||||
RUNNERS_REGISTRATION_ROOT="${RUNNER_REGISTRATION_ROOT:-/config/act_runner/reg}"
|
||||
RUNNERS_CACHE_CONFIG=""
|
||||
if [ -n "$RUNNER_CACHE_HOST" ] && [ -n "$RUNNER_CACHE_PORT" ] && [ -n "$RUNNER_CACHE_SECRET" ]; then
|
||||
if [ -n "$RUNNERS_CONFIG_BASE" ] && [ -f "$RUNNERS_CONFIG_BASE" ]; then
|
||||
__log "Base config: $RUNNERS_CONFIG_BASE"
|
||||
elif [ -n "$RUNNER_CACHE_HOST" ] && [ -n "$RUNNER_CACHE_PORT" ] && [ -n "$RUNNER_CACHE_SECRET" ]; then
|
||||
RUNNERS_CONFIG_BASE=""
|
||||
RUNNERS_CACHE_CONFIG="/config/act_runner/runners-cache.yaml"
|
||||
cat <<EOF >"$RUNNERS_CACHE_CONFIG"
|
||||
runner:
|
||||
file: REPLACE_RUNNER_STATE/.runner
|
||||
cache:
|
||||
enabled: true
|
||||
external_server: 'http://${RUNNER_CACHE_HOST}:${RUNNER_CACHE_PORT}/'
|
||||
secret: '${RUNNER_CACHE_SECRET}'
|
||||
EOF
|
||||
else
|
||||
RUNNERS_CONFIG_BASE=""
|
||||
fi
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
# Create the config used by one runner, isolating its registration state.
|
||||
__runner_config() {
|
||||
local runner_dir=$1
|
||||
local runner_config="$runner_dir/config.yaml"
|
||||
if [ -n "$RUNNERS_CONFIG_BASE" ]; then
|
||||
sed "s|^\([[:space:]]*\)file:.*|\1file: $runner_dir/.runner|" "$RUNNERS_CONFIG_BASE" >"$runner_config" || return 1
|
||||
echo "$runner_config"
|
||||
elif [ -n "$RUNNERS_CACHE_CONFIG" ]; then
|
||||
sed "s|REPLACE_RUNNER_STATE|$runner_dir|" "$RUNNERS_CACHE_CONFIG" >"$runner_config" || return 1
|
||||
echo "$runner_config"
|
||||
fi
|
||||
}
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
# Register a single runner synchronously (no daemon start)
|
||||
__register_runner() {
|
||||
local runner_id=$1
|
||||
local runner_name="${RUNNER_NAME_PREFIX:-runner}-${runner_id}"
|
||||
local runner_dir="/config/act_runner/reg/${runner_name}"
|
||||
local runner_dir="$RUNNERS_REGISTRATION_ROOT/${runner_name}"
|
||||
local config_args=()
|
||||
local runner_config=""
|
||||
|
||||
mkdir -p "$runner_dir"
|
||||
[ -d "$runner_dir" ] && cd "$runner_dir" || return 1
|
||||
[ -n "$RUNNERS_CACHE_CONFIG" ] && config_args=(--config "$RUNNERS_CACHE_CONFIG")
|
||||
runner_config="$(__runner_config "$runner_dir")" || return 1
|
||||
[ -n "$runner_config" ] && config_args=(--config "$runner_config")
|
||||
|
||||
if [ ! -f "$runner_dir/.runner" ]; then
|
||||
__log "Registering runner: $runner_name (ID: $runner_id)"
|
||||
@@ -139,11 +193,13 @@ __register_runner() {
|
||||
__start_runner_daemon() {
|
||||
local runner_id=$1
|
||||
local runner_name="${RUNNER_NAME_PREFIX:-runner}-${runner_id}"
|
||||
local runner_dir="/config/act_runner/reg/${runner_name}"
|
||||
local runner_dir="$RUNNERS_REGISTRATION_ROOT/${runner_name}"
|
||||
local config_args=()
|
||||
local runner_config=""
|
||||
|
||||
[ -d "$runner_dir" ] && cd "$runner_dir" || return 1
|
||||
[ -n "$RUNNERS_CACHE_CONFIG" ] && config_args=(--config "$RUNNERS_CACHE_CONFIG")
|
||||
runner_config="$(__runner_config "$runner_dir")" || return 1
|
||||
[ -n "$runner_config" ] && config_args=(--config "$runner_config")
|
||||
__log "Starting daemon for runner: $runner_name"
|
||||
exec act_runner daemon "${config_args[@]}"
|
||||
}
|
||||
@@ -155,20 +211,28 @@ done
|
||||
unset i
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
# Phase 2: start all daemons in parallel
|
||||
runner_pids=()
|
||||
for i in $(seq 1 $RUNNERS_START); do
|
||||
(__start_runner_daemon "$i") &
|
||||
runner_pids+=("$!")
|
||||
done
|
||||
unset i
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
RUNNERS_ID="$(jobs -p | tr '\n' ' ')"
|
||||
if [ -n "$RUNNERS_ID" ]; then
|
||||
__log "All $RUNNERS_START runners started successfully"
|
||||
__log "Process IDs: $RUNNERS_ID"
|
||||
printf 'All %s runners started successfully: %s\n' "$RUNNERS_START" "$RUNNERS_ID"
|
||||
else
|
||||
__log "The runners have failed to start"
|
||||
printf '%s\n' "The runners have failed to start"
|
||||
RUNNERS_ID="${runner_pids[*]}"
|
||||
__log "All $RUNNERS_START runners launched"
|
||||
__log "Process IDs: $RUNNERS_ID"
|
||||
printf 'All %s runners launched: %s\n' "$RUNNERS_START" "$RUNNERS_ID"
|
||||
# Wait for all background processes and preserve a failure status. A daemon that
|
||||
# exits is reported as failed instead of being treated as successfully started.
|
||||
runner_failures=0
|
||||
for runner_pid in "${runner_pids[@]}"; do
|
||||
if ! wait "$runner_pid"; then
|
||||
runner_failures=$((runner_failures + 1))
|
||||
__log "ERROR: Runner process $runner_pid exited unsuccessfully"
|
||||
fi
|
||||
done
|
||||
if [ "$runner_failures" -gt 0 ]; then
|
||||
__log "$runner_failures runner process(es) failed"
|
||||
printf '%s\n' "$runner_failures runner process(es) failed" >&2
|
||||
exit 1
|
||||
fi
|
||||
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
# Wait for all background processes
|
||||
wait
|
||||
Reference in new issue
Block a user