rpuch commented on code in PR #5323:
URL: https://github.com/apache/ignite-3/pull/5323#discussion_r1976989478
##########
modules/partition-replicator/src/main/java/org/apache/ignite/internal/partition/replicator/handlers/WriteIntentSwitchRequestHandler.java:
##########
@@ -117,12 +117,8 @@ public CompletableFuture<ReplicaResult>
handle(WriteIntentSwitchReplicaRequest r
.map(tableId ->
invokeTableWriteIntentSwitchReplicaRequest(tableId, request,
clockService.current(), senderId))
.collect(toList());
- // We choose current() to try to avoid compaction of the chosen
version (and we make sure it's not below commitTs [if it's a commit]
- // or txBeginTs [if it's an abort]). But there seems to be no
guarantee that the compactor will not remove this version.
- // TODO: IGNITE-24574 Introduce a mechanism to save the chosen catalog
version from being compacted too early.
Review Comment:
The comment was suggested to introduce a mechanism that would allow to hold
catalog compaction, but it turned out that we can solve the problem at hand a
lot easier. `reliableCatalogVersion` is used to make sure the follower has *at
least* the version we used when producing the command, so it might exceed the
'exact' value. So, if the 'exact' value was already compacted away, it's not a
problem, we can take any value that exceeds it (this is what I do by taking the
earliest version of the Catalog currently available).
I added a comment to the Jira issue explaining the change of plans.
--
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: notifications-unsubscr...@ignite.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
