[
https://issues.apache.org/jira/browse/KAFKA-18377?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alyssa Huang updated KAFKA-18377:
---------------------------------
Description:
Previously, this ticket was for investigating if we can short circuit to
unattached on receiving all endquorum acks.
Resigned state keeps track of `unackedVoters` and on pollResigned only sends
endQuorum requests to those voters. We thought it might be useful to short
circuit this to allow the replica to discover any new leader more quickly (say
in the case that the node was removed from quorum, the new leader would not be
sending beginquorum requests to this node). However, there is a flaw in that
logic - resigned must increase its epoch when transitioning, so any replica
that communicates with this replica would then have to bump its epoch.
We want to achieve a balance of allowing other replicas to make progress in
their elections and potentially become leader, while not having resigned
replicas sit doing nothing for too long.
was:
Resigned state keeps track of `unackedVoters` and on pollResigned only sends
endQuorum requests to those voters. Right now it is not used to short circuit
the transition to Unattached once the set is empty.
It would be useful to short circuit this to allow the replica to discover any
new leader more quickly (say in the case that the node was removed from quorum,
the new leader would not be sending beginquorum requests to this node).
> Investigate transition out of resigned state
> --------------------------------------------
>
> Key: KAFKA-18377
> URL: https://issues.apache.org/jira/browse/KAFKA-18377
> Project: Kafka
> Issue Type: Improvement
> Reporter: Alyssa Huang
> Assignee: Alyssa Huang
> Priority: Major
>
> Previously, this ticket was for investigating if we can short circuit to
> unattached on receiving all endquorum acks.
> Resigned state keeps track of `unackedVoters` and on pollResigned only sends
> endQuorum requests to those voters. We thought it might be useful to short
> circuit this to allow the replica to discover any new leader more quickly
> (say in the case that the node was removed from quorum, the new leader would
> not be sending beginquorum requests to this node). However, there is a flaw
> in that logic - resigned must increase its epoch when transitioning, so any
> replica that communicates with this replica would then have to bump its epoch.
>
> We want to achieve a balance of allowing other replicas to make progress in
> their elections and potentially become leader, while not having resigned
> replicas sit doing nothing for too long.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
