peterxcli opened a new pull request, #11403: URL: https://github.com/apache/ozone/pull/11403
## What changes were proposed in this pull request? `SCMContainerPlacementRackAware` can put a new replica on the rack of an existing replica. It picks random nodes on the first such rack and gives up after 3 unusable picks. Decommissioned and in-maintenance nodes stay in the network topology, so when that rack has only such nodes left, placement fails without trying the other replicas' racks or the fallback. Re-replication then stalls, and so does decommission. Now each of those racks gets its own retries, and after them placement moves on to the next rack, then to the fallback. The error after the last retry now says "No writable datanode with enough space was found", since unwritable nodes cause it too, not only full ones. Found with the SCM simulation (HDDS-16627). ## What is the link to the Apache JIRA https://issues.apache.org/jira/browse/HDDS-16628 ## How was this patch tested? New unit tests in `TestSCMContainerPlacementRackAware`. -- 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]
