[
https://issues.apache.org/jira/browse/KAFKA-19022?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17938478#comment-17938478
]
Lorcan edited comment on KAFKA-19022 at 3/26/25 8:20 AM:
---------------------------------------------------------
Hi [~rsamudrala], I've checked the KafkaRaftClient class for the
Inconsistent_Cluster_Id error and it looks like there are several instances
where a customised error message with the cluster ids is not being set:
VoteResponseData
BeginQuorumEpochResponseData
EndQuorumEpochResponseData
FetchResponseData
FetchSnapshotResponseData
These all implement the ApiMessage, just like AddRaftVoterResponseData does. It
looks like this would require a new schema for the above classes, given the
logic when new fields have been added and I'm not sure if this would require a
KIP or not.
was (Author: JIRAUSER308669):
Hi [~rsamudrala], I've checked the KafkaRaftClient class for the
Inconsistent_Cluster_Id error and it looks like there are several instances
where a customised error message with the cluster ids is not being set:
VoteResponseData
BeginQuorumEpochResponseData
EndQuorumEpochResponseData
FetchResponseData
FetchSnapshotResponseData
These all implement the ApiMessage, just like AddRaftVoterResponseData does. I
can create a PR to add an errorMessage field to FetchResponseData to get
feedback and see if this is a viable approach.
> Display cluster IDs being compared when encountering INCONSISTENT_CLUSTER_ID
> error
> ----------------------------------------------------------------------------------
>
> Key: KAFKA-19022
> URL: https://issues.apache.org/jira/browse/KAFKA-19022
> Project: Kafka
> Issue Type: Improvement
> Components: kraft, logging
> Affects Versions: 3.9.0
> Reporter: Ranganath Samudrala
> Assignee: Lorcan
> Priority: Major
>
> While migrating Kafka from zookeeper to kraft, we see errors in logs like
> {{INCONSISTENT_CLUSTER_ID in FETCH response }}
> or
> {{INCONSISTENT_CLUSTER_ID in VOTER response }}
> But cluster IDs being compared is not displayed in logs so there is not
> enough information to see where the issue is. Is the class data *clusterId*
> empty (which could potentially be a bug?) or incoming *clusterId* empty or
> incorrect?
> [KafkaRaftClient|https://github.com/apache/kafka/blob/31e1a57c41cf9cb600751669dc71bcd9596b45f9/raft/src/main/java/org/apache/kafka/raft/KafkaRaftClient.java#L1459]
> {code:java}
> private boolean hasValidClusterId(String requestClusterId) {
> // We don't enforce the cluster id if it is not provided.
> if (requestClusterId == null)Unknown macro: {
> return true;
> }
> return clusterId.equals(requestClusterId);
> }
> .
> .
> private CompletableFuture<FetchResponseData> handleFetchRequest(
> RaftRequest.Inbound requestMetadata,
> long currentTimeMs
> ) {
> FetchRequestData request = (FetchRequestData) requestMetadata.data();
> if (!hasValidClusterId(request.clusterId())) {
> return completedFuture(new
> FetchResponseData().setErrorCode(Errors.INCONSISTENT_CLUSTER_ID.code()));
> }
> .
> .
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
