smjn commented on code in PR #18848:
URL: https://github.com/apache/kafka/pull/18848#discussion_r1954506706
##########
group-coordinator/src/main/java/org/apache/kafka/coordinator/group/GroupCoordinatorService.java:
##########
@@ -822,21 +834,60 @@ public
CompletableFuture<DeleteGroupsResponseData.DeletableGroupResultCollection
});
groupsByTopicPartition.forEach((topicPartition, groupList) -> {
-
CompletableFuture<DeleteGroupsResponseData.DeletableGroupResultCollection>
future =
- runtime.scheduleWriteOperation(
- "delete-groups",
- topicPartition,
- Duration.ofMillis(config.offsetCommitTimeoutMs()),
- coordinator -> coordinator.deleteGroups(context, groupList)
- ).exceptionally(exception -> handleOperationException(
- "delete-groups",
- groupList,
- exception,
- (error, __) ->
DeleteGroupsRequest.getErrorResultCollection(groupList, error),
- log
- ));
-
- futures.add(future);
+
CompletableFuture<DeleteGroupsResponseData.DeletableGroupResultCollection>
shareFuture = deleteShareGroups(topicPartition, groupList)
Review Comment:
It was discouraged - we can do it ...
So, first we will send a list of all groups to GroupCoordinatorShard which
will return a list of delete requests only for VALID SHARE GROUPS. Other groups
will be ignored. This is the only extra work - we are only reading state from
the a hashmap.
We can alleviate the situation by checking `shareGroup.isEmpty()` in
`GroupMetadataManager`. That way, the `persister.deleteState` will not be
called for that group but the standard deleteGroups code will be called.
##########
group-coordinator/src/main/java/org/apache/kafka/coordinator/group/GroupCoordinatorService.java:
##########
@@ -822,21 +834,60 @@ public
CompletableFuture<DeleteGroupsResponseData.DeletableGroupResultCollection
});
groupsByTopicPartition.forEach((topicPartition, groupList) -> {
-
CompletableFuture<DeleteGroupsResponseData.DeletableGroupResultCollection>
future =
- runtime.scheduleWriteOperation(
- "delete-groups",
- topicPartition,
- Duration.ofMillis(config.offsetCommitTimeoutMs()),
- coordinator -> coordinator.deleteGroups(context, groupList)
- ).exceptionally(exception -> handleOperationException(
- "delete-groups",
- groupList,
- exception,
- (error, __) ->
DeleteGroupsRequest.getErrorResultCollection(groupList, error),
- log
- ));
-
- futures.add(future);
+
CompletableFuture<DeleteGroupsResponseData.DeletableGroupResultCollection>
shareFuture = deleteShareGroups(topicPartition, groupList)
Review Comment:
using persister in shard was discouraged - we can do it ...
So, first we will send a list of all groups to GroupCoordinatorShard which
will return a list of delete requests only for VALID SHARE GROUPS. Other groups
will be ignored. This is the only extra work - we are only reading state from
the a hashmap.
We can alleviate the situation by checking `shareGroup.isEmpty()` in
`GroupMetadataManager`. That way, the `persister.deleteState` will not be
called for that group but the standard deleteGroups code will be called.
--
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]
