Repository navigation
[CELEBORN-2482] Report pushes failed by push-timeout channel close as push timeout instead of connection exception - #3876
Open
afterincomparableyum wants to merge 1 commit into
Conversation
… push timeout instead of connection exception CELEBORN-2458 closes the channel when a push request times out. The other pushes still outstanding on that channel are then failed by channelInactive with "Connection from ... closed" (or by the write listener with "Failed to send request ..."), which ExceptionUtils.connectFail maps to PUSH_DATA_CONNECTION_EXCEPTION_PRIMARY/REPLICA. Those reports overwrite the worker's PUSH_DATA_TIMEOUT_* entry in the excluded workers, misattributing a worker that stopped responding as a connection failure, and make PushDataTimeoutTest "pushdata timeout will add to pushExcludedWorkers" flaky. Record the push timeout status before the timeout handler closes the channel and use it for the remaining pushes failed on that channel.
This branch has not been deployed
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What changes were proposed in this pull request?
When a push request times out and its timeout handler closes the channel (CELEBORN-2458),
TransportResponseHandlernow records the timeout status (PUSH_DATA_TIMEOUT_PRIMARY/REPLICA) before the close. The other pushes on that channel that fail because of the close are reported with this status instead of as connection exceptions. That covers bothchannelInactiveand write failures on the closed channel.Why are the changes needed?
Since CELEBORN-2458, a push timeout closes the channel. The other pushes still outstanding on it are then failed with
Connection from ... closed(orFailed to send request ...), whichExceptionUtils.connectFailmaps toPUSH_DATA_CONNECTION_EXCEPTION_PRIMARY/REPLICA. These collateral failures overwrite the worker'sPUSH_DATA_TIMEOUT_*entry in the excluded workers, so a worker that stopped responding gets misreported as a connection failure. This also makesPushDataTimeoutTestflaky:In the failed run, a push to a worker timed out (
PUSH_DATA_TIMEOUT_PRIMARY). In the same millisecond,Still have 1 requests outstanding when connection from ... is closedfailed a second push on that channel, and the worker was then reported asPUSH_DATA_CONNECTION_EXCEPTION_PRIMARY.Does this PR resolve a correctness bug?
Does this PR introduce any user-facing change?
How was this patch tested?
TransportResponseHandlerSuiteJ. The new push-timeout test fails without this change.TransportResponseHandlerSuiteJ(10 tests) andTransportClientFactorySuiteJ(11 tests): passed.PushDataTimeoutTestwith Spark 3.5: 5 tests passed.