poorbarcode commented on code in PR #23049:
URL: https://github.com/apache/pulsar/pull/23049#discussion_r1682361462
##########
pulsar-broker/src/main/java/org/apache/pulsar/broker/namespace/NamespaceService.java:
##########
@@ -1509,6 +1514,31 @@ public CompletableFuture<List<String>>
getListOfTopics(NamespaceName namespaceNa
}
}
+ public CompletableFuture<List<String>> getListOfUserTopics(NamespaceName
namespaceName, Mode mode) {
+ String key = String.format("%s://%s", mode, namespaceName);
+ CompletableFuture<List<String>> queryRes =
+ inProgressQueryUserTopics.computeIfAbsent(key, k -> {
+ CompletableFuture<List<String>> res = new CompletableFuture<>();
+ // Switch thread to avoid blocking other threads who are calling
the current method.
Review Comment:
> It seems that you're talking about TopicName#get could cost much time if
the cache has many (e.g. 10k) entries. However, costing time does not mean
"blocking". Moving this operation to another thread (pulsar.getExecutor()) does
not change anything except switching the CPU pressure from the current thread
to another thread, as well as a thread switch cost.
The code you provided
[here](https://github.com/apache/pulsar/pull/23049#discussion_r1682295880) is
the evidence of the first one task blocks other threads
--
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]
