* fix: Fail fast on Redis for cache and rate limits, lock migrations in Postgres
* fix: Use the fail-fast Redis client for locks
* fix: Require a database pool of at least two connections
* chore: Rename cache Redis client to ephemeralClient
* perf: Bound MutexLock retries and de-dupe same-key cache misses
MutexLock configured redlock with retryCount: 120, retryDelay: 1000 — a
2-minute retry window for locks with a 4s default TTL. Under cache-miss
storms (e.g. websocket reconnects all missing user.collectionIds) N
contenders piled onto one lock key, retaining ~K²/2 error objects plus
pending timers/promises. Bound retries to ~the lock TTL (retryCount: 15,
retryDelay: 250) and add a per-process in-flight de-dupe in
CacheHelper.getDataOrSet so N same-key misses share one lock acquisition
and callback.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* fix: Return falsy cached values instead of treating them as misses
getDataOrSet checked `if (cache)`, so a cached null/false/0/"" was treated
as a miss and re-populated on every read (e.g. Collection.getCachedDocument
Structure caches NavigationNode[] | null via the AfterSave hook). Check for
undefined — the only value getData returns on a genuine miss — so falsy
cached values are returned without an extra lock and DB query.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* deslop
* fix: Scope bounded lock retries to cache locks only
The tightened retry window (15×250ms) was set on the shared Redlock
instance, so it also affected long-lived callers — notably the 10-minute
"migrations" lock in checkMigrations, which fatal-exits on acquire failure
and would crash concurrent boots. Restore the generous default on the
shared instance and pass a bounded retry override (MutexLock.cacheRetry
Settings) only from the cache path via a new per-acquire retry option.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* fix: Cache falsy callback results in getDataOrSet
populate only cached truthy callback results, so a valid null (e.g. from
Collection.getCachedDocumentStructure) was never stored and every miss
re-ran the lock + DB query. Skip caching only when the result is undefined,
and guard the CacheResult detection against null (typeof null === "object").
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* perf: Use SCAN and batched UNLINK instead of KEYS in CacheHelper.clearData
The KEYS command blocks the Redis event loop while it walks the entire
keyspace, which stalls all other clients on large installations —
notably on startup where the whole unfurl cache prefix is cleared.
Replace it with incremental SCAN iteration and delete the matched keys
in batches with UNLINK, which reclaims memory asynchronously, so
neither discovery nor deletion blocks the server. Adds test coverage
for CacheHelper including multi-page prefix clears.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015K9AAwaGTLVQZnp7uv5Ebj
* fix: Address review comments on CacheHelper.clearData
Unlink each SCAN batch as the scan progresses instead of buffering all
matching keys in memory, repeating until a full pass matches nothing so
implementations with unstable scan cursors still clear every key. Wrap
the operation in try/catch with logging for consistency with the other
CacheHelper methods, and mark constants as readonly.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015K9AAwaGTLVQZnp7uv5Ebj
---------
Co-authored-by: Claude <noreply@anthropic.com>
* fix: Suspended users should not be included in cached member count for groups
* fix: Defer CounterCache hook registration until model is initialized
The previous test-only no-op hid a timing bug where setImmediate could
fire before the Sequelize instance had registered the related model,
causing "Model not initialized" failures. Poll until the model is
ready, and unref the pending immediate so it does not keep the event
loop alive in environments where the database is never initialized.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* perf: Reduce overhead of group member count invalidation
Select only the groupId column with raw queries and de-duplicate before
issuing Redis deletes, avoiding loading full GroupUser rows into memory
when a user belongs to many groups.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* chore: unref Redis healthcheck interval
Don't keep the Node event loop alive solely for the periodic ping; the
event loop should drain on its own when the application is shutting
down or a Jest worker is finishing.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* refactor: Centralize counter cache key in RedisPrefixHelper
Avoid duplicating the "count:<Model>:<relation>:<id>" string between
the CounterCache decorator and the User suspension hook by routing
both through a single getCounterCacheKey helper.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* fix: Walk to parent transaction when scheduling cache invalidation
Nested savepoints commit independently of their outer transaction, so
afterCommit callbacks attached to the inner transaction may run after
the outer rolls back, or never run at all. Match the pattern used in
Collection, Event, and base/Model and walk to the parent transaction
so the cache invalidation fires after the real outer commit.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
* chore: Address code quality findings
* Round 2, quality findings
* fix: Add fallback for MediaQueryList.addEventListener in test env
The jsdom test environment doesn't implement addEventListener on
MediaQueryList. Prefer addEventListener but fall back to the
deprecated addListener when unavailable.