fix Intermittent test failures in ProxyPublishConsumeTest.socketTest #253
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
merlimat
reviewed
Feb 28, 2017
| } catch (Exception e) { | ||
| log.error(e.getMessage()); | ||
| } | ||
| newFixedThreadPool(1).submit(() -> { |
There was a problem hiding this comment.
The executor thread will be leaked in this case
rdhabalia
force-pushed
the
socket_test
branch
2 times, most recently
from
cea3e06 to
1988a30
Compare
merlimat
reviewed
Mar 1, 2017
| } | ||
| }); | ||
| // let's wait for clients to be stopped | ||
| Thread.sleep(2000); |
There was a problem hiding this comment.
Instead of always sleeping 2 seconds, we could use the future returned by executor.submit() and do a get() with 2sec timeout.
Also, can you do the same change in ProxyPublishConsumeWithoutZKTest and ProxyPublishConsumeTls ?
sijie
pushed a commit
to sijie/pulsar
that referenced
this pull request
Mar 4, 2018
hangc0276
pushed a commit
to hangc0276/pulsar
that referenced
this pull request
May 26, 2021
Fixes apache#247 The original implementation of message publishing cannot guarantee that a topic partition's pending writes are completed in order. This PR is to fix the issue by refactoring the handler for Produce requests. The basic steps of handling a single partition's produce request are: Get PersistentTopic from topic manager; Convert Kafka's MemoryRecords to Pulsar's ByteBuf; Call PersistentTopic#publishMessages to write ByteBuf to BK asynchronously. This PR adds a PendingProduce class to compose pending step 1 and 2 to a CompletableFuture, then add the PendingProduce object to a queue (PendingProduceQueue) which is associated with the partition name in a map named pendingProduceQueueMap. When the step 1 and 2 are completed, the queue will try to remove all ready PendingProduce objects in the head and call PersistentTopic#publishMessages. We use synchronized keyword to make the remove operation thread safe. Therefore, the first two steps can be executed in parallel, but the third step is executed in order. In addition, this PR fixes the current message order test by using different batch size. In the original test, all messages are batched to a single batch, so the disorder never happens because there's only one batch. * Refactor handleProduceRequest to fix message disorder issue * Fix existed message order test * Synchronize PendingProduce#publishMessages * Test message order with different batch.size config
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation
#237 sometime
WebSocketClient.stop()gets stuck and it cause the test-case failure. So, wait for 2-seconds for clients to be closed and then ignore it.