HDFS-16184. De-flake TestBlockScanner#testSkipRecentAccessFile #3329
Conversation
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
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
Can you update the description with the reason for failure and detail about the fix |
|
Done @ayushtkn. Thanks |
ayushtkn
approved these changes
Aug 25, 2021
There was a problem hiding this comment.
Thanx @virajjasani for the fix, I tried reproducing this with a simple sleep, and got the same exception, and with the your fix that got sorted.
Changes LGTM
Will wait some time for Takanobu before pushing this.
|
Thanks @ayushtkn for detailed review with cross-verification. |
tasanuma
approved these changes
Aug 25, 2021
|
Also, it's good to see two consecutive +1s from QA bot for hadoop-hdfs. Haven't seen more than single +1 from QA build for long time on trunk. |
|
Merged. Thanks for your contribution, @virajjasani. Thanks for your review, @ayushtkn.
That's true. QA results seem pretty good recently. Really thanks for fixing several flaky tests, @virajjasani. |
tasanuma
pushed a commit
that referenced
this pull request
Aug 25, 2021
Reviewed-by: Ayush Saxena <ayushsaxena@apache.org> Signed-off-by: Takanobu Asanuma <tasanuma@apache.org> (cherry picked from commit 1b9927a)
Thanks @tasanuma for all the reviews!! |
kiran-maturi
pushed a commit
to kiran-maturi/hadoop
that referenced
this pull request
Nov 24, 2021
…e#3329) Reviewed-by: Ayush Saxena <ayushsaxena@apache.org> Signed-off-by: Takanobu Asanuma <tasanuma@apache.org>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
Description of PR
Test TestBlockScanner#testSkipRecentAccessFile is flaky:
Reason for failure:
Thread in VolumnScanner keeps scanning blocks. Before block scan, it executes
TestScanResultHandler#setupand it keeps waiting oninfoobject until main thread intestSkipRecentAccessFile()notifiesinfoobject and sets it'sshouldRunto true for VolumnScanner thread to successfully return fromTestScanResultHandler#setup. Now in main test, we try to assert thatinfo.blocksScannedstays 0 which is only possible if VolumnScanner's block scanner thread does not reachinfo.blocksScanned++point, which cannot be guaranteed always and hence this test is flaky.Fix:
In order to fix this, the main thread in
testSkipRecentAccessFile()should initializeinfo.semas Semaphore (permitting only single thread to take a lock, Mutex) and take a lock by main thread so that until released, block scan thread cannot reach pointinfo.blocksScanned++and hence the test never fails. Before the block scanner thread reach the point of incrementing blocksScanned count, it has to acquire Semaphore and it gets blocked here until main thread releases the lock.How was this patch tested?
Unit tests