jsancio commented on code in PR #15671:
URL: https://github.com/apache/kafka/pull/15671#discussion_r1586526016
##########
raft/src/main/java/org/apache/kafka/raft/ReplicatedLog.java:
##########
@@ -261,7 +260,7 @@ default long truncateToEndOffset(OffsetAndEpoch endOffset) {
* @param snapshotId the end offset and epoch that identifies the snapshot
* @return a writable snapshot if it doesn't already exist
*/
- Optional<RawSnapshotWriter> storeSnapshot(OffsetAndEpoch snapshotId);
+ Optional<RawSnapshotWriter> createNewSnapshotUnchecked(OffsetAndEpoch
snapshotId);
Review Comment:
So the use cases are very different hence the reason why I separated the
methods.
`createNewSnapshot` is indirectly called by the KRaft state machine. In this
case the `ReplicatedLog` makes sure that the provided snapshot id (offset and
epoch) is consistent with the log.
`createNewSnapshotUnchecked` is used by the local replica to override the
entire log with the leader's snapshot. For example, right after downloading the
snapshot we fully truncate the log:
https://github.com/apache/kafka/blob/trunk/raft/src/main/java/org/apache/kafka/raft/KafkaRaftClient.java#L1459-L1473
This part of the code calls `truncateToLatestSnapshot` which deletes all of
the old snapshots and replaces the log with an empty log that start at the just
downloaded snapshot (the latest snapshot). In other words
`truncateToLatestSnapshot` makes the `ReplicadLog` consistent after the direct
use of `createNewSnapshotUnchecked`.
--
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: jira-unsubscr...@kafka.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
