[jira] [Created] (KAFKA-21228) The cold snapshot timer should check pending updates instead of snapshot timestamps

["Chia-Ping Tsai (Jira)" <ji...@apache.org>]

Chia-Ping Tsai created KAFKA-21228:
--------------------------------------

             Summary: The cold snapshot timer should check pending updates instead of snapshot timestamps
                 Key: KAFKA-21228
                 URL: https://issues.apache.org/jira/browse/KAFKA-21228
             Project: Kafka
          Issue Type: Improvement
            Reporter: Chia-Ping Tsai
            Assignee: Chia-Ping Tsai


long coldSnapshottedPartitionsCount = shareStateMap.values().stream()
            .filter(shareGroupOffset -> shareGroupOffset.createTimestamp() - shareGroupOffset.writeTimestamp() != 0)
            .count();

        // If all share partitions are snapshotted, it means that
        // system is quiet and cold snapshotting will not help much.
        if (coldSnapshottedPartitionsCount == shareStateMap.size()) {
            log.debug("All share snapshot records already cold snapshotted, skipping.");
            return new CoordinatorResult<>(List.of(), null);
        }

 

The early return treats "all partitions cold-snapshotted" as "system is quiet". But ShareUpdate does not touch createTimestamp or writeTimestamp (see merge()), so once every partition has been cold-snapshotted, the timer returns early on every run until some partition reaches the records threshold. Slow-traffic partitions are never flushed, which conflicts with the purpose of the timer.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)