- Remove dead DOCKER_TIMEOUT=10 from stability_watchdog (daemon check
is delegated to docker_watchdog; variable was never used post-refactor)
- Fix parity check detection: parity-date.txt is gone in Unraid 7.3;
now reads mdResync from var.ini (non-zero = check/sync in progress),
with parity-date.txt fallback for older versions
- Same is_parity_running() fix in docker_watchdog
- DOCKER_TIMEOUT in docker_watchdog now reads WATCHDOG_DAEMON_TIMEOUT
from master.conf (default 20s) instead of hardcoded 15s
- WATCHDOG_DAEMON_RESTART_WAIT now reads from master.conf
(default 900s/15min) instead of hardcoded 30s
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
common.sh: add is_docker_enabled() and is_vm_manager_enabled() helpers
reading /boot/config/docker.cfg and /boot/config/domain.cfg.
docker_watchdog: exit cleanly if Docker not enabled in Unraid settings.
stability_watchdog: skip Docker daemon check and Docker container stop
if Docker not enabled; skip virsh VM shutdown if VM Manager not enabled.
server_reboot: skip VM shutdown and libvirt stop if VM Manager not enabled.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
docker restart uses the image ID baked in at container creation — it never
picks up a new digest. rebuild_container reads the stored Unraid XML template,
stops the old container, recreates it with the new image, then handles the
old image cleanup. This is what the Unraid UI 'Apply Update' button does.
docker_update.sh: add rebuild section after pull loop for DAILY containers.
docker_update_remaining.sh: replace retry_docker docker restart with rebuild_container.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
docker_update_remaining.sh already pruned; docker_update.sh (daily)
did not — orphaned images accumulated with every daily update run.
Same prune pattern as docker_update_remaining.sh.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Redis warns and can fail background saves without this. Applied via
inotify_tuning.sh which already handles kernel parameter tuning at
array start. Resets on each reboot — reapplied automatically.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Skip list belongs to docker_watchdog, not stability_watchdog (legacy
naming from when docker+system were the only two watchdogs). Rename
variable and physical file system_watchdog_failed.db →
docker_watchdog_failed.db across all files.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
stability_watchdog: remove SYS_WATCHDOG_FAILED_FILE from state file
touch list — skip list is owned by docker_watchdog, not stability.
docker_watchdog: rename check_system_watchdog_state →
check_resource_watchdog_state (reads RW_STATE_FILE from resource_watchdog,
not SYS_WATCHDOG_STATE_FILE from stability_watchdog). Update all comments
and state file docs to reflect correct ownership.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Previously only cleared inside the strikes>0 block. After a reboot,
strikes reset to 0 but the flag persists on flash — daemon could be
healthy but flag stays true, causing stability_watchdog to keep striking.
Now cleared on every healthy docker info check regardless of strike count.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
docker_watchdog now writes daemon_confirmed_down=true to its state file
when restart is attempted and daemon is still unresponsive. Clears the
flag on recovery.
stability_watchdog removes the duplicate CRITICAL daemon check (which
bypassed all strikes, all abort conditions, and had its own rc.docker
restart). Replaced with a standard strike check reading the
daemon_confirmed_down flag from docker_watchdog's state file.
Result: docker_watchdog owns all daemon restart logic. stability_watchdog
escalates to reboot only after N consecutive strike cycles confirming the
daemon is truly unrecoverable — same pattern as all other standard checks.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
stability_watchdog: container health is owned by docker_watchdog — rebooting
when docker_watchdog skip-lists a container creates a reboot loop. Removed
SYS_WATCHDOG_CHECK_CONTAINERS check entirely. docker_watchdog handles
strikes, restart attempts, skip-listing, and notifications.
varaverk.plg: warn on USB/removable boot device at install time. Plugin is
designed for internal NVMe/SSD boot. Warning is non-blocking but clear.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Exits cleanly before startup grace check if /mnt/user is not mounted
as shfs. Prevents watchdogs running against Docker/storage that isn't
available, and blocks false-positive reboots when array is stopped.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
runc v1.3.5 (Docker 29.x / Unraid 7.3+) crashes on bind-propagation=shared
with any container — not GPU-specific. Ramdisk tmpfs is already MS_SHARED
at kernel level; propagation is automatic and the option is unnecessary.
- Remove bind-propagation=shared from all example Extra Parameters
- Update "Why shared Is Required" → "Mount Propagation — Unraid 7.3+"
- Update GPU troubleshooting section to cover the general runc crash
- Verify commands updated (no longer checking for Propagation: shared)
- Historical notes preserved explaining why it existed before
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
runc v1.3.5 (Docker 29.x / Unraid 7.3+) rejects bind-propagation=shared
on symlink sources with: open /proc/self/mountinfo: no such file or directory.
The ramdisk tmpfs is already MS_SHARED at the kernel level — propagation is
inherited automatically and does not need to be specified in Docker.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
228 path references updated from /mnt/user/appdata/Varaverk/ to
/boot/config/plugins/varaverk/. PATHS header block updated to match.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Plugin folder /boot/config/plugins/varaverk is the single home for
scripts, data, and state. Available at boot before array mounts.
USB boot is not a supported configuration.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Bootstrap varaverk.cfg default and conf template default back to
/mnt/user/appdata/Varaverk — safe for USB boot users. Flash paths
(/boot/config/plugins/varaverk) are set in the user's gitignored
personal configs and don't affect public installs.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Move SCRIPTS_DIR/DATA_DIR/STATE_DIR from appdata to /boot/config/plugins/varaverk
- All state files now in STATE_DIR (no more /tmp or /boot/config root writes)
- Bootstrap: Gitea-first clone with GitHub fallback, no array dependency
- varaverk.cfg seeded with Gitea connection settings
- .gitignore: add State_Files/, varaverk.cfg, varaverk-*.txz
- Partnership/transcode/fallback scripts use STATE_DIR variables
- PHP config.php: DATA_DIR/STATE_DIR constants, VV_SETUP_STATE_FILE dynamic
- deploy.sh PROD_ROOT updated to plugin flash dir
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Was looking for host1.conf at SCRIPTS_ROOT/host1.conf instead of
SCRIPTS_ROOT/Configurations/host1.conf — causing the 'host1.conf not found'
warning and leaving HOST1_SSH_KEY unwritten after key generation.
flag_key was declared with 'local' at script top level, which is only
valid inside bash functions. This left flag_key empty, writing '=true'
(malformed) to setup.db instead of 'HOST1_LOCAL_DONE=true'.
Changed to a plain variable assignment.
- Single '▶ Onboard' button per unprovisioned remote → expands that card
to show Step 1 (terminal link + click-to-copy command) and Step 2 (Push Conf)
- Onboard state persists across 10s poll re-renders via top-level _vvOnboarding map
and _vvPtReload hook set by the IIFE
- Phase 1 complete: amber 'SSH ready' pill + Tailscale IP + Phase 2 manual button
- Phase 2 complete: green 'Active' pill + date + IP + uptime + unRAID version + state
- Compact compact '○ Not provisioned' + Reset button when not in onboard mode
- Removed vvPtPhase1 (replaced by vvPtStartOnboard toggle)
- vvPtStartOnboard / vvPtEndOnboard are top-level functions
At phase 0, show below the buttons:
- 'Open Terminal' link using HOST1's Tailscale IP (/webterminal/ttyd/)
- Clickable command code block (click to copy) with the Phase 1 script path
- Note that tab auto-updates on next poll when the script completes
Tailscale IP comes from selfNode.ts_ip already in the API response.
partnership_onboard.sh (--phase1-only):
- Test if SSH already works first (BatchMode) — skip ssh-copy-id if yes
- If ssh-copy-id fails (no TTY in background): soft-fail, generate key locally,
print pubkey for manual install, write HOST2_KEY_READY=true, run local setup
- Only skip conf push if SSH truly unavailable; otherwise proceeds normally
- Exit 0 with clear manual-step instructions instead of hard failure
include/partnership.php:
- Read HOST2_KEY_READY from setup.db; expose as key_ready on remote nodes
pages/partnership.php:
- Node card: amber "Key generated · install on HOST2 then Push Conf" badge
when key_ready=true and phase=0
- "Push Conf (key installed)" button appears when key_ready=true — runs
--phase1-only --skip-ssh (assumes user manually did ssh-copy-id)
- vvPtPushConf() top-level function
vv_partner_state() now reads setup.db and adds onboard_phase (0/1/2/null)
to each host entry.
Partner card JS renders a badge under the hostname for remote hosts that
haven't completed Phase 2: amber "Awaiting onboard" (phase 1) or grey
"Not provisioned" (phase 0). Phase 2 complete shows nothing extra.
When HOST1 completes its local setup (--onboard --local-only), enable the
partnership in master.conf so sync scripts start functioning immediately.
Pushes the updated master.conf to all listed hosts via vv_push_master_conf().
partnership_manager.sh:
- --local-only flag for --onboard: skips remote pre-flight, WebUI reconfig,
state push; runs FolderView3 folder creation + writes HOST1_LOCAL_DONE=true
to setup.db + pushes it; exits without writing ACTIVE (partnership not yet
established until Phase 2)
partnership_onboard.sh:
- Phase 1 exit now runs partnership_manager.sh --onboard --local-only after
SSH + conf push, so HOST1 is fully configured on its own side before Phase 2
include/partnership.php:
- vv_pt_nodes() reads HOST1_LOCAL_DONE from setup.db; exposes as local_done
on the self node (null for remote nodes)
pages/partnership.php:
- Self node card shows green "Local setup complete · waiting for partner"
badge when local_done=true
- Actions: "Complete HOST1 Setup" button appears when HOST1 hasn't run local
setup yet but HOST2 is already at phase 1 — lets user trigger it manually
- vvPtLocalSetup() top-level function calls partnership_manager.sh via run.php
If run.php returns an empty body (transient nginx/PHP-FPM issue), the script
still runs via nohup. Treat empty response as ok instead of throwing a
JSON parse error that misleads the user into thinking the action failed.
onboard_cancel.sh:
- SSHes HOST2 while key still valid: removes HOST1 pubkey from authorized_keys
and clears HOST2_PHASE* from setup.db on the remote side
- Deletes local private + public key pair
- Clears HOST2_PHASE* from local setup.db
- Graceful if HOST2 unreachable (local side still cleaned up)
partnership.php:
- Cancel button (warn style, slightly dimmed) at phase 0 and phase 1
- vvPtCancel() top-level function, calls onboard_cancel.sh via run.php
window.fn = inside an IIFE is not reliable in all script-loading contexts.
Move vvPtPhase1/2, vvPtOnboard, vvPtOffboard and their _vvPtRun helper
above the IIFE as plain top-level function declarations — unconditionally
global and reachable from onclick= attributes.
Tab pages use full page reloads (not AJAX) — function declarations inside
the IIFE are not in global scope, so onclick= attributes fail silently.
Fix: assign all onclick-callable functions to window.*
Replace vvRunById() calls with direct fetch (vvRunById is also IIFE-scoped
in scheduler.php, unavailable in the partnership tab context).
Add .catch() handlers so network/parse errors surface as alerts.
include/monitor.php: vv_fallback_state() now returns:
enabled, check_interval, handback_strikes, handback_strikes_required,
partnership_suspended, partner_lost_at, partnership_required,
partnership_suspend_after — all read from state file + conf vars.
pages/monitor.php: fallback card render rebuilt:
Disabled: clear message with how to enable
Suspended: partnership inactive with elapsed time
FAILOVER: outage duration + T1/T2/T3/T4 tier badges (active=green,
pending=dark) + handback strike progress dots (●●○) when remote
comes back and strikes are counting
NO_INTERNET/DARK: short contextual line
NORMAL: quiet "monitoring · 30s · partnership gated" meta line
Accent border: disabled/suspended → warn, NORMAL → ok, FAILOVER → err
End users leave GITEA_REPO_PATH="" in master.conf (template default).
Before hitting any Gitea detection logic, the script now checks for this
and does git pull --ff-only origin main directly — origin is already
GitHub since the .plg cloned from there.
Sparse checkout skipped on this path: conf files are gitignored so
nothing sensitive is in the repo to exclude.
Developer path (GITEA_REPO_PATH configured) is completely unchanged.
unraid_api.php: track key_missing separately from general fallbacks.
vv_api_data() sets key_missing=true when HOST*_UNRAID_API_KEY is empty.
vv_api_get_status() includes key_missing in the status payload.
monitor.php: two distinct banner states instead of one orange warning.
key_missing → subtle dark/grey note: "API key not configured — add
HOST1_UNRAID_API_KEY in Scheduler → host conf to enable enhanced monitoring"
key present but unreachable → existing orange ⚠ with fallback list.
No banner at all when API is working correctly.
Plugin/varaverk.plg:
- Add support="https://github.com/FailedProxy/Varaverk/issues"
- Install completion log now says "open Varaverk to complete setup"
with navigation hint — user lands on wizard automatically
Plugin/unraid/VaraverkSettings.page:
- Fix stale SCRIPTS_DIR default (unraid_scripts → Varaverk)
- Add GitHub / Issues / Changelog links at bottom of settings page
Plugin/unraid/Varaverk.page:
- Small "⎋ GitHub" link right-aligned in the tab bar
links to github.com/FailedProxy/Varaverk in a new tab
After Step 8 (arr bootstrap), owner calls vv_push_master_conf() and
vv_push_setup_state() via php -r. SSH is established at this point so
the probe passes and all listed hosts get the authoritative master.conf
immediately — no manual save required.
Skipped if onboard failed, dry-run mode, or php unavailable (with
fallback instruction). Result shown in onboard summary as Step 9.
Fix 1 — Monitor partner card (common.php + monitor.php):
vv_remote_hosts_stats() no longer skips hosts with no API key.
Returns no_api_key:true entry instead. Monitor JS renders
"API key not configured — complete Onboard to enable" instead
of a blank space.
Fix 2 — rawconf push not alarming pre-onboard (config.php + scheduler.php):
Probe failures (Tailscale not found, plugin not installed) now carry
ready:false. JS treats ready:false results silently — button shows
"✓ Saved" not "push failed: HOST2" before SSH is set up.
Only genuine post-onboard failures (ready not false) show as errors.
Fix 3 — Partnership Onboard button (partnership.php):
Disabled with tooltip when no partner hostname is in master.conf.
When partner configured but not onboarded, adds explanation note:
"Onboard will generate your SSH key, exchange it with the partner..."
No partner configured: "Edit master.conf and set HOST2."
