patsonluk commented on code in PR #909:
URL: https://github.com/apache/solr/pull/909#discussion_r922387781
##########
solr/solrj/src/java/org/apache/solr/common/cloud/ZkStateReader.java:
##########
@@ -539,17 +672,17 @@ private void constructState(Set<String>
changedCollections) {
if (log.isDebugEnabled()) {
log.debug(
"clusterStateSet: interesting [{}] watched [{}] lazy [{}] total
[{}]",
- collectionWatches.keySet().size(),
- watchedCollectionStates.keySet().size(),
+ collectionWatches.watchedCollections().size(),
+ collectionWatches.activeCollectionCount(),
lazyCollectionStates.keySet().size(),
clusterState.getCollectionStates().size());
}
if (log.isTraceEnabled()) {
log.trace(
"clusterStateSet: interesting [{}] watched [{}] lazy [{}] total
[{}]",
- collectionWatches.keySet(),
- watchedCollectionStates.keySet(),
+ collectionWatches.watchedCollections(),
+ collectionWatches.activeCollections(),
lazyCollectionStates.keySet(),
clusterState.getCollectionStates());
}
Review Comment:
I believe it could change but it should be okay.
As far as I understand, we only need a strong guarantee/synchronization on
the write ops. As for read op, it appears none of the read operation requires
such guarantee, ie even if the value read has changed or is inconsistent with
the `watchedCollectionStates`, it should not cause any serious issues. (it
might print value that does not reflect the latest set or fetch/not fetch
something - but these are not new behaviors)
The core of this PR is to remove the race conditions between write
operations of 2 separate maps, which can cause stale state. And now since all
write ops are on the same map with sync, we should be safe 😊 . As for read
ops, it's pretty much the same as before, which does not need to have such
strong guarantee I suppose?
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
