- Bump utopia-php/database from 5.3.17 to 5.3.19
- Remove invalid (int) cast on tenant sequence in shared tables mode
- Fix DSN construction for documentsdb/vectorsdb: filter empty strings
from explode(), skip pool filtering when shared tables env vars unset,
fail fast when no pool found
- Use dedicated mode for separate database pools in getDatabasesDB
since shared tables can't work across engines (PostgreSQL integer
_tenant vs MongoDB UUID tenant). Auto-init schema on first use.
- Add documentsdb/vectorsdb shared tables env vars to CI workflow
- Fix testChannelTablesDBRowUpdate race condition with deterministic
event drain loop
The timeout parameters were accidentally removed in daaf1d7f77, causing
the assertEventually to use the default timeout which is too short for
deployment builds in CI.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The root cause (PostgreSQL duplicate index race condition) has been
fixed in the database library, so the timeout workarounds are no
longer needed.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The assertEventually call inside Coroutine\run() causes a fatal PHP
error (exit 255) when it times out, killing the entire test process.
Moving DB/collection/attribute setup and polling outside the coroutine
ensures PHPUnit can properly handle assertion failures.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
PostgreSQL adapter throws non-Duplicate exceptions when a table/index
already exists during concurrent createCollection. Catch Throwable and
attempt metadata-only creation as fallback. Also increase Realtime
coroutine attribute polling timeout from 30s to 60s.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Realtime E2E tests were taking 24+ minutes due to intentional timeout waits. Many tests verify event filtering by expecting a TimeoutException, and each was waiting 45 seconds.
Changes:
- Reduce default WebSocket timeout from 45s to 2s in RealtimeBase
- Add optional timeout parameter to getWebsocket() methods
- Use longer timeouts (5-10s) for tests that legitimately wait for slow operations (function executions, test channel events)
- Use named parameters for improved readability
Performance impact:
- Before: 24:07 minutes (1,447 seconds)
- After: 1:09 minutes (69 seconds)
- Speedup: ~21x faster
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Realtime: Use getLastEmailByAddress with userId probe instead of getLastEmail
to prevent email race conditions in parallel execution; add status assertions
on verification/recovery PUT responses for clearer failure diagnostics
- Avatars: Wrap external URL image fetches in assertEventually to retry
transient network failures from appwrite.io
- Migrations: Increase performMigrationSync timeout from 10s to 60s
- GraphQL: Increase attribute polling timeouts from 60s to 120s with 500ms
intervals to accommodate PostgreSQL under CI load
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix float/int/bool attribute creation failures caused by type loss
during queue serialization in the Databases worker
- Rework session limit test to retry session creation for cache
propagation in shared mode
- Increase GraphQL attribute polling timeouts from 30s to 60s
- Increase SchemaPolling waitForAllAttributes timeout to 180s
- Increase Realtime WebSocket client timeout from 30s to 45s
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Parallel test execution via ParaTest creates more queue load than a single
database worker can handle, causing attributes to get stuck at 'processing'.
Increase _APP_WORKERS_NUM to 4 to match parallelism, increase WebSocket
timeout from 15s to 30s, and add retry logic for proxy requests in Sites tests.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Increase SchemaPolling default timeouts from 8min to 15min for
waitForAttribute, waitForAttributes, waitForAttributeCount, and
waitForIndex to match waitForAllAttributes/waitForAllIndexes. Under
parallel test load the database worker gets backlogged, causing
attributes to stay in 'processing' state longer than 8 minutes.
- Add isEmpty check in Functions Delete before updating schedules to
prevent 500 errors when the schedule document is missing (consistent
with other modules like Executions/Delete).
- Add retry logic to cleanupFunction in FunctionsBase for transient
MongoDB transaction abort errors during function deletion.
- Increase Realtime attribute/index polling timeouts from 2min to 15min
to handle worker backlog under CI load.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add --exclude-group ciIgnore to paratest commands in tests.yml
(the @group ciIgnore annotation was defined but never excluded)
- Increase attribute/index polling timeouts in RealtimeConsoleClientTest
from 15s to 120s for Shared V2 stability
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Move maxLength/assertStringContainsString from testUpdateAttributeEnum
back to testPatchAttribute where $attribute is defined
- Fix remaining $data['moviesId'] -> $moviesId in testCreateAttributes
- Reduce WebSocket timeout from 30s to 15s for faster failure detection
- Reduce Realtime helper assertEventually from 30s to 15s
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
DatabasesBase:
- testCreateAttributes: use local $moviesId/$actorsId instead of $data['...']
- testValidateOperators: add missing setupOneToManyRelationship() call
- testSelectQueries: add missing setupOneToManyRelationship() call
- testUniqueIndexDuplicate: define missing $serverHeaders variable
- testListDocuments: change return type from array to void (no callers)
RealtimeConsoleClientTest:
- testAttributesCollectionsAPI: create database/collection before opening
WebSocket to prevent their events from interfering with attribute events
- testAttributesTablesAPI: same restructuring for TablesDB variant
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The TablesDB index creation API requires 'columns' parameter, not
'attributes' (which is the collections API parameter). This caused
testDeleteIndexTablesAPI to always fail because the index was never
actually created.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add response status checks on all creation calls (database, table,
column, index) to catch silent failures early and provide diagnostic
info instead of mysterious 404s during polling.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The Realtime test helpers used fragile sleep(5) for attribute/index
creation and had short polling timeouts (30s/60s) that fail under CI
parallel load. Replaced with assertEventually polling and increased
all timeouts to 120s.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Matej Bačo
Parth BhardwajandGitHub
Prem Palanisamy
ArnabChatterjee20kandGitHub
ArnabChatterjee20k
Jake BarnbyandClaude Opus 4.6
loks0nandClaude Sonnet 4.5
copilot-swe-agent[bot]andabnegate