PrefectHQ
prefect
Blog
Docs
Changelog
Blog
Docs
Changelog
Overview
Branches
Benchmarks
Runs
Performance History
Latest Results
Make UI V2 the default for self-hosted servers (#22519) Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: Alex S <alex.s@prefect.io> Co-authored-by: alex.s <ajstreed1@gmail.com> Co-authored-by: Alex Streed <desertaxle@users.noreply.github.com>
main
1 minute ago
Keep default heartbeat retention override; index alone fixes the full scan The missing event_resources.event_id index is what caused the retention- override vacuum to full-scan. With that index the per-type path is cheap, so keep the default heartbeat 7d override, which still caps high-volume heartbeat events at 7d even when server.events.retention_period is longer. Co-authored-by: Alexander Streed <alex.s@prefect.io>
devin1/oss-8084-db_vacuum-retention-override-delete-full-scans
3 minutes ago
fix: reduce default lease_duration and add timeout to concurrency context manager (closes #16299)
botbikamordehai2-sketch:fix/issue-16299-1784548434
2 hours ago
Use camel case in facet json Signed-off-by: merobi-hub <merobi@gmail.com>
merobi-hub:add-openlineage
3 hours ago
Add index on event_resources(event_id) to fix full-scan in retention-override vacuum closes #22535 The vacuum_events_with_retention_overrides task deletes event_resources rows using an event_id IN (subquery) pattern, but EventResource had no index on event_id, causing a full sequential scan on large tables. This consistently timed out (default 10s) and silently broke event retention. Changes: 1. ORM: Add ix_event_resources__event_id index to EventResource 2. Migration (PostgreSQL): CREATE INDEX CONCURRENTLY for zero-downtime 3. Migration (SQLite): CREATE INDEX (compatible with SQLite) 4. Query: Add EventResource.occurred pre-filter so the query can use both the existing occurred index and the new event_id index
Kunj157:fix/add-event-resources-event-id-index
3 hours ago
Add event_resources.event_id index and drop redundant heartbeat retention override The db_vacuum retention-override task deletes event_resources via an `event_id IN (SELECT ...)` subquery, but event_resources had no index on event_id, so every batch sequentially scanned the whole table, timed out under the API statement timeout, and deleted nothing. - Add an index on event_resources(event_id) (ORM + SQLite/Postgres migrations) - Default event_retention_overrides to empty; the previous {"prefect.flow-run.heartbeat": 7d} default matched the global 7d events retention, so it only added a full scan while duplicating work already done by the indexed vacuum_old_events path Closes #22535 Co-authored-by: Alexander Streed <alex.s@prefect.io>
devin1/oss-8084-db_vacuum-retention-override-delete-full-scans
6 hours ago
Scope orchestration client reuse to the actions consumer Address review feedback: instead of a process-level cache of shared clients keyed by settings identity, have the actions consumer open one OrchestrationClient and one PrefectServerEventsAPIClient for its lifetime and close them on shutdown. Actions reuse them via contextvars; when no consumer is active (e.g. a direct call in a test) the accessors fall back to a short-lived client. Per-action automation headers remain request-local through the existing scoped_headers() contextvar + httpx request hook, so concurrent actions sharing a client never clobber each other's headers. Removes the _SharedClientMixin (cache, lock, shared(), _reset_shared()) and the autouse test fixture that reset it. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
fatih-acar:share-orchestration-clients
6 hours ago
Emit flow events once and preserve hyphenated task names Signed-off-by: merobi-hub <merobi@gmail.com>
merobi-hub:add-openlineage
17 hours ago
Latest Branches
CodSpeed Performance Gauge
0%
Fix db_vacuum retention-override full scan of event_resources
#22536
16 minutes ago
8936f44
devin1/oss-8084-db_vacuum-retention-override-delete-full-scans
CodSpeed Performance Gauge
0%
fix: reduce default lease_duration and add timeout to concurrency context manager (closes #16299)
#22538
2 hours ago
4dca34f
botbikamordehai2-sketch:fix/issue-16299-1784548434
CodSpeed Performance Gauge
0%
Add OpenLineage integration
#22475
3 hours ago
1e8fca9
merobi-hub:add-openlineage
© 2026 CodSpeed Technology
Home
Terms
Privacy
Docs