created transcode manager

This commit is contained in:
2026-04-11 08:34:00 -04:00
parent 44915887c2
commit 70a58feee6
4 changed files with 829 additions and 39 deletions
+61 -39
View File
@@ -36,6 +36,9 @@
# MEDIA PERMISSIONS Share list, mode and owner for permissions script
# MEDIA CLEANER Anime and media folder lists and file patterns
#
# ── TRANSCODES ─────────────────────────────────────────────────────────────────────────────
# TRANSCODE MANAGER Ramdisk and SSD fallback transcode management
#
# ── SYSTEM WATCHDOG ────────────────────────────────────────────────────────────────────────
# SYSTEM WATCHDOG System health monitoring — last line of defense
#
@@ -90,7 +93,6 @@
# ━━━ Daily Sync Shares ━━━
# Shares synced once daily by Orchestrators/daily_sync.sh
# No profile needed — all fall through to DEFAULT_RSYNC_OPTS above.
# Add or remove paths here to manage what gets synced.
DAILY_SYNC_SHARES=(
/mnt/user/Anime_Movies-Old
/mnt/user/Anime_Shows-Old
@@ -227,6 +229,7 @@ WEEKLY_RESTART_CONTAINERS=(
# 8GB=8192 6GB=6144 4GB=4096 2GB=2048 1GB=1024
declare -A WATCHDOG_CONTAINERS=(
["Emby"]=16384
["jellyfin_with_request"]=12288
["LidaTube"]=6144
["Tdarr"]=6144
["Code-Server"]=1024
@@ -235,6 +238,7 @@ declare -A WATCHDOG_CONTAINERS=(
# Containers to check HTTP responsiveness — omit to skip
declare -A WATCHDOG_CONTAINER_URLS=(
["Emby"]="http://localhost:8096"
["Jellyfin-Gmer4Lfe"]="http://localhost:8095"
)
# Containers that should always be running — monitored for unexpected stops
@@ -250,7 +254,7 @@ WATCHDOG_REQUIRED_CONTAINERS=(
# CPU thresholds — normalised against total core count at runtime
SOFT_CPU_THRESHOLD=80 # warn at this % of total system CPU
HARD_CPU_THRESHOLD=90 # strike at this % of total system CPU
HARD_CPU_THRESHOLD=85 # strike at this % of total system CPU
CPU_FAIL_LIMIT=2 # consecutive strikes before restart
# Memory threshold
@@ -265,24 +269,19 @@ WATCHDOG_REQUIRED_CONTAINERS=(
# ==============================================================================================
# ━━━ Reboot ━━━
# User warning delay before scheduled reboot (seconds)
REBOOT_SLEEP=300
REBOOT_SLEEP=300 # user warning delay before scheduled reboot (seconds)
# ━━━ Mover ━━━
# Timeout before stopping the mover (seconds)
MOVER_STOP_TIMEOUT=300
MOVER_STOP_TIMEOUT=300 # timeout before stopping the mover (seconds)
# ━━━ Syslog Filter ━━━
# Path for the rsyslog Docker noise filter file
FILTER_FILE="/etc/rsyslog.d/ignore-docker-veth.conf"
# ━━━ PHP-FPM ━━━
# PHP-FPM config file path and max children value
PHP_CONF="/etc/php-fpm.d/www.conf"
PHP_MAX_CHILDREN=250
# ━━━ Clear Logs ━━━
# System log files to clear on each run
LOG_FILES=(/var/log/syslog /var/log/messages /var/log/dmesg)
# ==============================================================================================
@@ -290,7 +289,6 @@ WATCHDOG_REQUIRED_CONTAINERS=(
# ==============================================================================================
# ━━━ Media Permissions ━━━
# Mode and owner applied recursively to all listed shares
PERMISSIONS_MODE="777"
PERMISSIONS_OWNER="nobody:users"
@@ -340,7 +338,6 @@ MEDIA_CLEAN_FOLDERS=(
/mnt/user/Tv_Shows
)
# Anime file patterns — junk files common in anime downloads
ANIME_FILE_PATTERNS=(
'*.sfv' '*.md5' '*.sha1' '*.txt' '*.url' '*.lnk'
'*.rar' '*.zip' '*.info' '*.torrent' '*.sample*' '*.proof*'
@@ -348,7 +345,6 @@ ANIME_FILE_PATTERNS=(
'*.log' '*.json'
)
# Media file patterns — includes *.iso and *.lrc not needed in anime
MEDIA_FILE_PATTERNS=(
'*.sfv' '*.md5' '*.sha1' '*.txt' '*.url' '*.lnk'
'*.rar' '*.zip' '*.info' '*.torrent' '*.sample*' '*.proof*'
@@ -356,44 +352,70 @@ MEDIA_FILE_PATTERNS=(
'*.log' '*.json' '*.iso' '*.lrc'
)
# ==============================================================================================
# ── TRANSCODES ────────────────────────────────────────────────────────────────────────────────
# ==============================================================================================
# Session-based storage allocator using filesystem indirection.
# New transcode sessions land wherever TRANSCODE_LINK points.
# Existing sessions are never interrupted — ffmpeg resolves path once at session start.
#
# Flow:
# ramdisk_setup.sh — run once at array start, creates ramdisk and symlink
# transcode_manager.sh — runs every 2-3 min, monitors usage and flips symlink
# transcode_cleanup.sh — runs every 5 min, removes old inactive files
#
# Hysteresis gap between RAMDISK_WARN_GB and RAMDISK_LOW_GB prevents flip-flop
# when usage hovers near the threshold. Gap should be at least 0.5-1GB.
# ━━━ Transcode Manager ━━━
# Paths
RAMDISK_PATH="/mnt/ramdisk_transcodes" # tmpfs mount point
RAMDISK_SIZE="8G" # increase if you have RAM headroom
TRANSCODE_LINK="/mnt/user/ram-transcode" # symlink Emby points at
TRANSCODE_SSD="/mnt/cache/Temp_Storage/Emby/Transcodes/" # SSD fallback location
# Thresholds in GB
RAMDISK_WARN_GB=6.8 # flip symlink to SSD at or above this usage
RAMDISK_LOW_GB=5.5 # flip symlink back to ramdisk when usage drops here
RAMDISK_SSD_MIN_GB=20 # minimum free GB on SSD before allowing flip — abort if below
# Cleanup settings
TRANSCODE_MAX_AGE=20 # minutes before a file is eligible for cleanup
TRANSCODE_ORPHAN_AGE=30 # minutes before an orphaned file is eligible (slightly longer buffer)
# Flip frequency monitoring
TRANSCODE_FLIP_WARN=3 # notify if symlink flips this many times in one hour
# Permissions — should match your Emby container user
TRANSCODE_OWNER="nobody:users"
TRANSCODE_MODE="755"
# ==============================================================================================
# ── SYSTEM WATCHDOG ───────────────────────────────────────────────────────────────────────────
# ==============================================================================================
# Last line of defense — reboots the system cleanly if it is about to become unstable.
# Designed to run on a cron schedule (recommended every 15-30 minutes).
# Works alongside docker_watchdog.sh — containers first, system second.
#
# Strike system — sustained threshold hits trigger reboot, not single spikes.
# Reboot loop protection — shuts down instead of rebooting if limit hit in window.
# Container skip list — persistent, auto-clears when container recovers.
# Thresholds set at "about to fall over" levels — not just high usage.
# ━━━ System Watchdog State Files ━━━
# Strike counts — /tmp resets on reboot, correct for strike tracking
SYS_WATCHDOG_STATE_FILE="/tmp/system_watchdog_state.db"
# Persistent container skip list — survives reboots, auto-clears on recovery
SYS_WATCHDOG_FAILED_FILE="/boot/config/system_watchdog_failed.db"
# Reboot timestamp log — survives reboots for loop detection
SYS_WATCHDOG_REBOOT_LOG="/boot/config/system_watchdog_reboots.db"
# ━━━ Strike and Reboot Loop Settings ━━━
# Consecutive threshold hits before triggering reboot
SYS_WATCHDOG_STRIKE_LIMIT=2
# Maximum reboots allowed within the window before shutdown instead
SYS_WATCHDOG_REBOOT_LIMIT=3
# Window in hours — controls both reboot count window AND rolling log purge
# 12 = entries older than 12hrs purge automatically / 24 = entries older than 24hrs purge
SYS_WATCHDOG_REBOOT_WINDOW_HRS=12
SYS_WATCHDOG_STRIKE_LIMIT=2 # consecutive hits before reboot trigger
SYS_WATCHDOG_REBOOT_LIMIT=3 # max reboots in window before shutdown instead
SYS_WATCHDOG_REBOOT_WINDOW_HRS=12 # window in hours — controls reboot count AND log purge
# ━━━ Thresholds ━━━
# Set at "about to fall over" levels — not just high usage
SYS_WATCHDOG_ROOTFS_PCT=95 # rootfs usage % before strike
SYS_WATCHDOG_LOG_PCT=95 # /var/log usage % before strike
SYS_WATCHDOG_MEM_GB=4 # free RAM in GB below which strikes
SYS_WATCHDOG_ROOTFS_PCT=95 # rootfs % before strike
SYS_WATCHDOG_LOG_PCT=95 # /var/log % before strike
SYS_WATCHDOG_MEM_GB=4 # free RAM GB below which strikes (128GB system)
SYS_WATCHDOG_ARC_PINNED_PCT=98 # ZFS ARC % of max before reclaim attempt
SYS_WATCHDOG_ARC_RELEASE_PCT=95 # ZFS ARC % after reclaim that still triggers
SYS_WATCHDOG_LOAD_MULTIPLIER=32 # strike if load avg > cores x this value
SYS_WATCHDOG_ZOMBIE_LIMIT=50 # strike if zombie process count exceeds this
SYS_WATCHDOG_CPU_TEMP_MAX=95 # degrees C — tjmax varies by CPU
SYS_WATCHDOG_LOAD_MULTIPLIER=3 # strike if load avg > cores x this value
SYS_WATCHDOG_ZOMBIE_LIMIT=50 # strike if zombie count exceeds this
SYS_WATCHDOG_CPU_TEMP_MAX=95 # degrees C — adjust for your CPU tjmax
# ━━━ Check Toggles ━━━
# true = run this check / false = skip entirely
@@ -402,17 +424,17 @@ MEDIA_FILE_PATTERNS=(
SYS_WATCHDOG_CHECK_RAM=true
SYS_WATCHDOG_CHECK_ARC=true
SYS_WATCHDOG_CHECK_CPU_TEMP=true
SYS_WATCHDOG_CHECK_LOAD=true
SYS_WATCHDOG_CHECK_LOAD=false
SYS_WATCHDOG_CHECK_ZOMBIES=true
SYS_WATCHDOG_CHECK_CONTAINERS=true
SYS_WATCHDOG_CHECK_DOCKER_DAEMON=true
# ━━━ Abort Toggles ━━━
# true = abort reboot if condition is active / false = reboot anyway
# Default true = conservative / set false to reboot regardless
# true = abort reboot if condition active / false = reboot anyway
# Default true = conservative set false only when you want "reboot no matter what"
SYS_WATCHDOG_ABORT_ON_ZFS_UNHEALTHY=true
SYS_WATCHDOG_ABORT_ON_PARITY=false
SYS_WATCHDOG_ABORT_ON_MOVER=false
SYS_WATCHDOG_ABORT_ON_PARITY=true
SYS_WATCHDOG_ABORT_ON_MOVER=true
# ==============================================================================================
# ──────────────────────── End Of User Variables ───────────────────────────────────────────────