Vinz2168 commented on code in PR #11885:
URL: https://github.com/apache/nifi/pull/11885#discussion_r4466594609
##########
nifi-framework-bundle/nifi-framework/nifi-web/nifi-web-api/src/main/java/org/apache/nifi/web/dao/impl/StandardComponentStateDAO.java:
##########
@@ -108,6 +112,9 @@ private void clearState(final String componentId, final
ComponentStateDTO compon
if (hasExactlyOneKeyRemoved(currentClusterState,
newClusterState)) {
manager.setState(newClusterState, Scope.CLUSTER);
Review Comment:
Good catch, thanks. This race was there before this PR, but it is closely
tied to this change, so I addressed it here in 45ebb60ef1. Both the cluster and
local paths now use `manager.replace()` with the `StateMap` that was read. When
the replace fails in a cluster, the DAO reads the state again. If it already
matches the requested state (another node applied the same replicated request
between the read and the replace), the request is a no-op. Otherwise it fails
with a conflict saying the state was modified concurrently, so a processor
update is never overwritten. I added unit tests for both interleavings, plus
local key removal with and without a concurrent update.
--
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]
