dataroaring commented on PR #60310: URL: https://github.com/apache/doris/pull/60310#issuecomment-3867556132
### Code review Found 1 issue: 1. Missing `has_mtime()` check when accessing `cluster.mtime()` in `get_cluster_status()`. On the server side (`meta_service_resource.cpp`), `mtime` is conditionally set only when `cluster.has_mtime()` is true. When it is not set, protobuf3 returns the default value `0`, so `0 * 1000 = 0` is stored as `mtime_ms` in the `ClusterStatusCache`. Downstream in `should_skip_compaction()`, the elapsed time is computed as `now - status_mtime`. When `status_mtime` is 0, `elapsed` equals the current Unix time in milliseconds (~1.7 trillion), which is always greater than the 5-minute takeover timeout. This causes immediate compaction takeover by non-writer clusters, bypassing the timeout safety mechanism entirely. https://github.com/apache/doris/blob/4d3741931e91c2c543f1ed8d076ff29efcc64a84/be/src/cloud/cloud_meta_mgr.cpp#L2281-L2283 Server-side conditional setting that proves `mtime` is optional: https://github.com/apache/doris/blob/4d3741931e91c2c543f1ed8d076ff29efcc64a84/cloud/src/meta-service/meta_service_resource.cpp#L4587-L4589 Downstream timeout bypass when `mtime_ms` is 0: https://github.com/apache/doris/blob/4d3741931e91c2c543f1ed8d076ff29efcc64a84/be/src/cloud/cloud_cluster_info.cpp#L166-L180 🤖 Generated with [Claude Code](https://claude.ai/code) <sub>- If this code review was useful, please react with 👍. Otherwise, react with 👎.</sub> -- 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]
