Chia-Ping Tsai created KAFKA-21362:
--------------------------------------
Summary: Classic group waits for the rebalance timeout after a
member is rejected with GROUP_MAX_SIZE_REACHED
Key: KAFKA-21362
URL: https://issues.apache.org/jira/browse/KAFKA-21362
Project: Kafka
Issue Type: Bug
Affects Versions: 4.2.2, 4.3.1, 4.2.1, 4.1.2, 4.1.1, 4.0.2, 4.3.0, 4.2.0,
4.1.0, 4.0.1, 4.0.0, 4.3.2, 4.4.0
Reporter: Chia-Ping Tsai
Assignee: Chia-Ping Tsai
This is a regression in the new coordinator.
The new coordinator does not try to complete the rebalance after removing the
member, so the rebalance can’t be completed in time.
{code:java}
.setMemberId(UNKNOWN_MEMBER_ID)
.setErrorCode(Errors.GROUP_MAX_SIZE_REACHED.code())
);
+// result = maybeCompleteJoinPhase(group); // one-line fix
} else if (isUnknownMember) {
result = classicGroupJoinNewMember(
context,
{code}
By contrast, the old coordinator always tries to complete the delayed
operation, so those classic consumers can start to fetch records soon.
{code:java}
// attempt to complete JoinGroup
if (group.is(PreparingRebalance)) {
rebalancePurgatory.checkAndComplete(GroupJoinKey(group.groupId))
}
{code}
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
