rick/LEDMatrix
1
0
Fork 0
mirror of https://github.com/ChuckBuilds/LEDMatrix.git synced 2026-04-11 13:23:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(plugins): stop reconciliation install loop, slow plugin list, and uninstall resurrection

Browse Source 
Three interacting bugs reported by a user (Discord/ericepe) on a fresh install:

1. The state reconciler retried failed auto-repairs on every HTTP request,
   pegging CPU and flooding logs with "Plugin not found in registry: github
   / youtube". Root cause: ``_run_startup_reconciliation`` reset
   ``_reconciliation_started`` to False on any unresolved inconsistency, so
   ``@app.before_request`` re-fired the entire pass on the next request.
   Fix: run reconciliation exactly once per process; cache per-plugin
   unrecoverable failures inside the reconciler so even an explicit
   re-trigger stays cheap; add a registry pre-check to skip the expensive
   GitHub fetch when we already know the plugin is missing; expose
   ``force=True`` on ``/plugins/state/reconcile`` so users can retry after
   fixing the underlying issue.

2. Uninstalling a plugin via the UI succeeded but the plugin reappeared.
   Root cause: a race between ``store_manager.uninstall_plugin`` (removes
   files) and ``cleanup_plugin_config`` (removes config entry) — if
   reconciliation fired in the gap it saw "config entry with no files" and
   reinstalled. Fix: reorder uninstall to clean config FIRST, drop a
   short-lived "recently uninstalled" tombstone on the store manager that
   the reconciler honors, and pass ``store_manager`` to the manual
   ``/plugins/state/reconcile`` endpoint (it was previously omitted, which
   silently disabled auto-repair entirely).

3. ``GET /plugins/installed`` was very slow on a Pi4 (UI hung on
   "connecting to display" for minutes, ~98% CPU). Root causes: per-request
   ``discover_plugins()`` + manifest re-read + four ``git`` subprocesses per
   plugin (``rev-parse``, ``--abbrev-ref``, ``config``, ``log``). Fix:
   mtime-gate ``discover_plugins()`` and drop the per-plugin manifest
   re-read in the endpoint; cache ``_get_local_git_info`` keyed on
   ``.git/HEAD`` mtime so subprocesses only run when the working copy
   actually moved; bump registry cache TTL from 5 to 15 minutes and fall
   back to stale cache on transient network failure.

Tests: 16 reconciliation cases (including 5 new ones covering the
unrecoverable cache, force-reconcile path, transient-failure handling, and
recently-uninstalled tombstone) and 8 new store_manager cache tests
covering tombstone TTL, git-info mtime cache hit/miss, and the registry
stale-cache fallback. All 24 pass; the broader 288-test suite continues to
pass with no new failures.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Chuck
2026-04-08 09:46:55 -04:00
parent 781224591f
commit b031b99130
6 changed files with 506 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

92
web_interface/blueprints/api_v3.py
View File

@@ -1714,9 +1714,23 @@ def get_installed_plugins():
import json
from pathlib import Path
# Re-discover plugins to ensure we have the latest list
# This handles cases where plugins are added/removed after app startup
api_v3.plugin_manager.discover_plugins()
# Re-discover plugins only if the plugins directory has actually
# changed since our last scan, or if the caller explicitly asked
# for a refresh. The previous unconditional ``discover_plugins()``
# call (plus a per-plugin manifest re-read) made this endpoint
# O(plugins) in disk I/O on every page refresh, which on an SD-card
# Pi4 with ~15 plugins was pegging the CPU and blocking the UI
# "connecting to display" spinner for minutes.
force_refresh = request.args.get('refresh', '').lower() in ('1', 'true', 'yes')
plugins_dir_path = Path(api_v3.plugin_manager.plugins_dir)
try:
current_mtime = plugins_dir_path.stat().st_mtime if plugins_dir_path.exists() else 0
except OSError:
current_mtime = 0
last_mtime = getattr(api_v3, '_installed_plugins_dir_mtime', None)
if force_refresh or last_mtime != current_mtime:
api_v3.plugin_manager.discover_plugins()
api_v3._installed_plugins_dir_mtime = current_mtime
# Get all installed plugin info from the plugin manager
all_plugin_info = api_v3.plugin_manager.get_all_plugin_info()
@@ -1729,17 +1743,10 @@ def get_installed_plugins():
for plugin_info in all_plugin_info:
plugin_id = plugin_info.get('id')
# Re-read manifest from disk to ensure we have the latest metadata
manifest_path = Path(api_v3.plugin_manager.plugins_dir) / plugin_id / "manifest.json"
if manifest_path.exists():
try:
with open(manifest_path, 'r', encoding='utf-8') as f:
fresh_manifest = json.load(f)
# Update plugin_info with fresh manifest data
plugin_info.update(fresh_manifest)
except Exception as e:
# If we can't read the fresh manifest, use the cached one
logger.warning("[PluginStore] Could not read fresh manifest for %s: %s", plugin_id, e)
# Note: we intentionally do NOT re-read manifest.json here.
# discover_plugins() above already reparses manifests on change;
# re-reading on every request added ~1 syscall+json.loads per
# plugin per request for no benefit.
# Get enabled status from config (source of truth)
# Read from config file first, fall back to plugin instance if config doesn't have the key
@@ -2369,14 +2376,25 @@ def reconcile_plugin_state():
from src.plugin_system.state_reconciliation import StateReconciliation
# Pass the store manager so auto-repair of missing-on-disk plugins
# can actually run. Previously this endpoint silently degraded to
# MANUAL_FIX_REQUIRED because store_manager was omitted.
reconciler = StateReconciliation(
state_manager=api_v3.plugin_state_manager,
config_manager=api_v3.config_manager,
plugin_manager=api_v3.plugin_manager,
plugins_dir=Path(api_v3.plugin_manager.plugins_dir)
plugins_dir=Path(api_v3.plugin_manager.plugins_dir),
store_manager=api_v3.plugin_store_manager,
)
result = reconciler.reconcile_state()
# Allow the caller to force a retry of previously-unrecoverable
# plugins (e.g. after the registry has been updated or a typo fixed).
force = False
if request.is_json:
payload = request.get_json(silent=True) or {}
force = bool(payload.get('force', False))
result = reconciler.reconcile_state(force=force)
return success_response(
data={
@@ -2822,6 +2840,22 @@ def uninstall_plugin():
if api_v3.operation_queue:
def uninstall_callback(operation):
"""Callback to execute plugin uninstallation."""
# Drop a tombstone *first* so a background reconciliation pass
# that interleaves with us cannot see the transient
# "config-entry-with-no-files" state and resurrect the
# plugin the user just asked to delete.
if hasattr(api_v3.plugin_store_manager, 'mark_recently_uninstalled'):
api_v3.plugin_store_manager.mark_recently_uninstalled(plugin_id)
# Clean up plugin configuration BEFORE removing files so the
# config entry and the on-disk files disappear together from
# the reconciler's point of view.
if not preserve_config:
try:
api_v3.config_manager.cleanup_plugin_config(plugin_id, remove_secrets=True)
except Exception as cleanup_err:
logger.warning("[PluginUninstall] Failed to cleanup config for %s: %s", plugin_id, cleanup_err)
# Unload the plugin first if it's loaded
if api_v3.plugin_manager and plugin_id in api_v3.plugin_manager.plugins:
api_v3.plugin_manager.unload_plugin(plugin_id)
@@ -2844,13 +2878,6 @@ def uninstall_plugin():
if api_v3.schema_manager:
api_v3.schema_manager.invalidate_cache(plugin_id)
# Clean up plugin configuration if not preserving
if not preserve_config:
try:
api_v3.config_manager.cleanup_plugin_config(plugin_id, remove_secrets=True)
except Exception as cleanup_err:
logger.warning("[PluginUninstall] Failed to cleanup config for %s: %s", plugin_id, cleanup_err)
# Remove from state manager
if api_v3.plugin_state_manager:
api_v3.plugin_state_manager.remove_plugin_state(plugin_id)
@@ -2879,6 +2906,18 @@ def uninstall_plugin():
)
else:
# Fallback to direct uninstall
# Tombstone + config cleanup happen BEFORE file removal — see
# queue path above for the full rationale (prevents reconciler
# resurrection race).
if hasattr(api_v3.plugin_store_manager, 'mark_recently_uninstalled'):
api_v3.plugin_store_manager.mark_recently_uninstalled(plugin_id)
if not preserve_config:
try:
api_v3.config_manager.cleanup_plugin_config(plugin_id, remove_secrets=True)
except Exception as cleanup_err:
logger.warning("[PluginUninstall] Failed to cleanup config for %s: %s", plugin_id, cleanup_err)
# Unload the plugin first if it's loaded
if api_v3.plugin_manager and plugin_id in api_v3.plugin_manager.plugins:
api_v3.plugin_manager.unload_plugin(plugin_id)
@@ -2891,13 +2930,6 @@ def uninstall_plugin():
if api_v3.schema_manager:
api_v3.schema_manager.invalidate_cache(plugin_id)
# Clean up plugin configuration if not preserving
if not preserve_config:
try:
api_v3.config_manager.cleanup_plugin_config(plugin_id, remove_secrets=True)
except Exception as cleanup_err:
logger.warning("[PluginUninstall] Failed to cleanup config for %s: %s", plugin_id, cleanup_err)
# Remove from state manager
if api_v3.plugin_state_manager:
api_v3.plugin_state_manager.remove_plugin_state(plugin_id)