epugh commented on PR #4857: URL: https://github.com/apache/solr/pull/4857#issuecomment-5530764741
TL;DR, things look good! ## Testing notes: disruptor 3.4.4 → 4.0.0 logging comparison Solr's default production logging config uses `<AsyncLogger>`/`<AsyncRoot>` (`solr/server/resources/log4j2.xml`), which is backed by LMAX Disruptor's ring buffer — so every log line already exercises this dependency. Built and ran both versions (`./gradlew :solr:packaging:dev`, `bin/solr start -e techproducts`), then hit each with an identical load: 40 concurrent workers × 50 iterations × 4 request types (match-all, OR query, a guaranteed-bad field lookup to trigger ERROR logging, and a two-field facet) — 8,000 requests in ~14s per run, ~14,100 log lines generated each. | Check | Result | |---|---| | ERROR lines (`undefined field`) | 2000 vs 2000 — exact match | | WARN lines | 7 vs 7 — exact match, identical message types | | INFO line count | 8110 vs 8130 (0.25% diff — normal run-to-run timing noise, not systematic) | | FATAL / dropped-event / ring-buffer-full / overflow / backpressure mentions | none in either log | | `disruptor` / `RingBuffer` mentions | none in either log | | Stop/shutdown behavior | clean on both, no timeout or hung-shutdown messages | No behavioral or logging differences observed between 3.4.4 and 4.0.0 under either idle startup or high-concurrency burst load. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
