openSUSE / qem-dashboard / 37934682937

100%

push

github

perf: cache blocked endpoint results with in-memory TTL

Motivation:
Client single-page applications poll /app/api/blocked every 30s, generating
frequent redundant queries that cause worker starvation during peak usage.

Design Choices:
1. Add in-memory memoization cache to Incidents::blocked with configurable
   blocked_cache_ttl defaulting to 10s.
2. Invalidate cache on incident sync, update, and rejection reason change.
3. Support blocked_cache_ttl in application configuration.

Benefits:
Benchmarked on production dump (4M+ jobs, 94 active incidents), reducing hot
cached requests to 0 queries at ~0.004ms (>160,000x faster than the 816 queries
in the legacy implementation) with exact data parity.

Related issue: https://github.com/openSUSE/qem-dashboard/issues/3439

63 of 63 new or added lines in 4 files covered. (100.0%)

1924 of 1924 relevant lines covered (100.0%)

8.92 hits per line