KAFKA-15003: When partitions are partition assignments change we do n… #13735
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ot update TopicIdReplicaAssignment during migration (DUAL_WRITE). Fixed the metadata change events in the Migration component to check correctly for the diff in existing topic changes and replicate the metadata to the Zookeeper. Also, made the diff check exhaustive enough to handle the partial writes in Zookeeper when we're try to replicate changes using a snapshot in the event of Controller failover. Add migration client and integration tests to verify the change.
mumrah
reviewed
May 22, 2023
Member
mumrah
left a comment
mumrah
left a comment
There was a problem hiding this comment.
This is a great catch @akhileshchg! Looks like we had some gaps in topic/partition metadata even beyond the snapshot handling.
Left some comments inline
metadata/src/main/java/org/apache/kafka/metadata/migration/KRaftMigrationZkWriter.java
Outdated
Show resolved
Hide resolved
metadata/src/main/java/org/apache/kafka/metadata/migration/KRaftMigrationZkWriter.java
Outdated
Show resolved
Hide resolved
metadata/src/main/java/org/apache/kafka/metadata/migration/KRaftMigrationZkWriter.java
Outdated
Show resolved
Hide resolved
metadata/src/main/java/org/apache/kafka/metadata/migration/KRaftMigrationZkWriter.java
Outdated
Show resolved
Hide resolved
metadata/src/main/java/org/apache/kafka/metadata/migration/KRaftMigrationZkWriter.java
Outdated
Show resolved
Hide resolved
metadata/src/main/java/org/apache/kafka/metadata/migration/KRaftMigrationZkWriter.java
Show resolved
Hide resolved
Comment on lines
164
to
165
| newPartitions.put(topicId, newTopicPartitions); | ||
| changedTopics.add(topicId); |
Member
There was a problem hiding this comment.
I think it's possible that we can see a difference between topicPartitionsInZk (computed based on the partition znodes) and the topic assignment in Kraft, but actually have a correct assignment in ZK (stored in the topic znode). That said, I don't think there's any harm in re-writing the kraft assignment to ZK.
cmccabe
reviewed
May 31, 2023
| val (migrationZkVersion, responses) = zkClient.retryMigrationRequestsUntilConnected(requests, state) | ||
| val resultCodes = responses.map { response => response.path -> response.resultCode }.toMap | ||
| if (resultCodes(TopicZNode.path(topicName)).equals(Code.NODEEXISTS)) { | ||
| // topic already created, just return |
Contributor
There was a problem hiding this comment.
should this read "partition already created"?
Contributor
|
Do we handle the case where # of partitions decreased? How do we handle changes to the partition znodes? Looks like we catch EEXIST in a bunch of cases, do we follow up to see that it has the right stuff in there? |
Contributor
Author
|
You can’t shrink number of partitions right? Except when the topic is
completely deleted? I think we handle that case.
On Wed, May 31, 2023 at 11:46 AM Colin Patrick McCabe < ***@***.***> wrote:
Do we handle the case where # of partitions decreased?
How do we handle changes to the partition znodes? Looks like we catch
EEXIST in a bunch of cases, do we follow up to see that it has the right
stuff in there?
—
Reply to this email directly, view it on GitHub
<#13735 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAPTMROW33NT2Q4GQ3RGOS3XI6GZNANCNFSM6AAAAAAYIB7K4U>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
--
Sent from my phone
|
Member
|
@akhileshchg there's a case where a topic is deleted and re-created with fewer partitions. Since ZK indexes topics by name, we can't readily see if the deletion happened when processing a KRaft snapshot. I added logic to handle the case of unexpected partitions in ZK. |
cmccabe
reviewed
Jun 1, 2023
| return builder.toString(); | ||
| } | ||
|
|
||
| public boolean hasSameAssignment(PartitionRegistration registration) { |
Contributor
There was a problem hiding this comment.
This should be equals. I'll fix it
cmccabe
pushed a commit
that referenced
this pull request
Jun 1, 2023
Fixed the metadata change events in the Migration component to check correctly for the diff in existing topic changes and replicate the metadata to the Zookeeper. Also, made the diff check exhaustive enough to handle the partial writes in Zookeeper when we're try to replicate changes using a snapshot in the event of Controller failover. Add migration client and integration tests to verify the change. Co-authored-by: Akhilesh Chaganti <[email protected]>
cmccabe
pushed a commit
that referenced
this pull request
Jun 1, 2023
Fixed the metadata change events in the Migration component to check correctly for the diff in existing topic changes and replicate the metadata to the Zookeeper. Also, made the diff check exhaustive enough to handle the partial writes in Zookeeper when we're try to replicate changes using a snapshot in the event of Controller failover. Add migration client and integration tests to verify the change. Co-authored-by: Akhilesh Chaganti <[email protected]>
Contributor
|
committed |
emissionnebula
added a commit
to confluentinc/kafka
that referenced
this pull request
Jun 16, 2023
…tream-3.5 * commit 'c2f6f29ca6e1306ac77ec726bac4cd09bd1aa80b': (76 commits) KAFKA-15019: Improve handling of broker heartbeat timeouts (apache#13759) KAFKA-15003: Fix ZK sync logic for partition assignments (apache#13735) MINOR: Add 3.5 upgrade steps for ZK and KRaft (apache#13792) KAFKA-15010 ZK migration failover support (apache#13758) KAFKA-15017 Fix snapshot load in dual write mode for ClientQuotas and SCRAM (apache#13757) MINOR: Update LICENSE-binary following snappy upgrade (apache#13791) Upgrade to snappy v1.1.10.0 (apache#13786) KAFKA-15004: Fix configuration dual-write during migration (apache#13767) KAFKA-8713: JsonConverter replace.null.with.default should prevent emitting default for Struct fields (apache#13781) KAFKA-14996: Handle overly large user operations on the kcontroller (apache#13742) ...
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
…ot update TopicIdReplicaAssignment during migration (DUAL_WRITE).
Fixed the metadata change events in the Migration component to check correctly for the diff in existing topic changes and replicate the metadata to the Zookeeper. Also, made the diff check exhaustive enough to handle the partial writes in Zookeeper when we're try to replicate changes using a snapshot in the event of Controller failover.
Add migration client and integration tests to verify the change.