paulononato/backstage
1
0
Fork 0
Code Issues Pull Requests Actions 50 Packages Projects Releases Wiki Activity

fix(catalog-backend): cache and cheapen catalog_entities_count metric

Browse Source 
The legacy Prometheus and OpenTelemetry observable gauges previously each
ran the per-kind count query against the search table on every metrics
scrape. With multiple pods and short scrape intervals, identical
sequential scans piled up faster than they completed, contending for
buffers in the database.

Extract a shared helper that wraps a 30-second TTL cache around a single
query, and have both gauges read from it. The query itself moves from
the (large) search table to final_entities, parsing kind out of
entity_ref via per-engine substring functions. The emitted labels and
values are unchanged.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Signed-off-by: Fredrik Adelöw <freben@spotify.com>
This commit is contained in:
Fredrik Adelöw
2026-05-08 11:48:58 +02:00
parent 4bb649d511
commit ccbad9d892
3 changed files with 223 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.changeset/catalog-metrics-cache.md
+9
View File
@@ -0,0 +1,9 @@
---
'@backstage/plugin-catalog-backend': patch
---
Improved the performance of the `catalog_entities_count` metric.
The legacy Prometheus and OpenTelemetry observable gauges previously each ran their own copy of the per-kind count query against the `search` table on every metrics scrape. On large catalogs this could pile up faster than the queries completed, contending for buffers and stalling the database.
The two callbacks now share a single query result with a short in-process TTL cache, and the underlying query reads from `final_entities` instead of `search`, avoiding the bitmap heap scans that dominated the previous form. The emitted labels and values are unchanged.