rkhachatryan commented on a change in pull request #18391:
URL: https://github.com/apache/flink/pull/18391#discussion_r794396173
##########
File path:
flink-state-backends/flink-statebackend-changelog/src/main/java/org/apache/flink/state/changelog/ChangelogStateBackend.java
##########
@@ -271,7 +273,7 @@ public StateBackend configure(ReadableConfig config,
ClassLoader classLoader)
keyedStateHandle instanceof
ChangelogStateBackendHandle
? (ChangelogStateBackendHandle)
keyedStateHandle
: new ChangelogStateBackendHandleImpl(
-
singletonList(keyedStateHandle),
+
singletonMap(UUID.randomUUID(), keyedStateHandle),
Review comment:
Maybe we should reconsider
https://github.com/apache/flink/pull/18391#discussion_r787631347?
> Btw, won't it solve the problem if we register all files in
IncrementalRemoteKeyedStateHandle with SharedStateRegistry (including metadata
etc.)? This would be much simpler I think in all respects.
> I guess previously there just was no need to register them (and they must
be kept separate for restore).
You said
> I think changelog keyed state backend should be designed to wrap any keyed
state backend
But plugging in a custom backend is different from plugging in a custom
Handle. The latter is currently not possible without patching Flink. We should
fix this at some point though :/
---
Another option is to disallow recovery from incremental non-changelog
checkpoints (leaving savepoints and full RocksDB checkpoints). But this seems
quite a big limitation.
--
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: issues-unsubscr...@flink.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
