shuke987 opened a new pull request, #65876: URL: https://github.com/apache/doris/pull/65876
## Summary Add bounded MetaService memory diagnostics to the Cloud P0 pipeline. - sample MetaService RSS, virtual memory, thread count, and host available memory every 5 seconds - trigger detailed diagnostics when RSS reaches 3 GiB or grows by at least 1 GiB in 60 seconds - capture `/proc/<pid>/status`, `smaps_rollup`, host memory, fd/task counts, and filtered brpc bvars - cap detailed snapshots and bvar output to avoid log growth - preserve Doris logs when diagnostics trigger, even if the regression run otherwise succeeds ## Motivation Several unrelated ASAN Cloud P0 runs showed MetaService RSS increasing by 4-18 GiB within one minute before host low-water failures and loss of port 5000. The existing 60-second system resource snapshots can miss the transition and do not retain enough MetaService-specific evidence to identify the allocator or subsystem. The lightweight 5-second samples provide the timeline at low volume. More expensive data is collected only after a threshold is crossed. ## Validation - `bash -n regression-test/pipeline/common/doris-utils.sh regression-test/pipeline/cloud_p0/run.sh` - `git diff --check` - synthetic absolute-RSS threshold fixture - synthetic growth-threshold fixture - detailed-snapshot limit and trigger-marker verification ## Check List - [x] No product behavior change - [x] Pipeline script validation - [ ] Cloud P0 validation -- 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]
