You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Test failure in System.Net.Http.Functional.Tests.SyncHttpHandler_HttpClientHandler_Cancellation_Test.GetAsync_CancelDuringResponseBodyReceived_Unbuffered_TaskCanceledQuickly #74122
Thanks @CarnaViire . Also please decide if it needs a fix to get backported to RC1, if it meets the bar. If it's a test being disabled, you can backport that too to clean the other PRs CI results.
cluster('engsrvprod.kusto.windows.net').database('engineeringdata').AzureDevOpsTests
| where TestName contains'GetAsync_CancelDuringResponseBodyReceived_Unbuffered_TaskCanceledQuickly'
| where RunCompleted > ago(10d)
@rzikm it looks like these work items ran during the storage outage we had a couple days ago and since we rely on uploading test data to our storage account to ingest into Kusto, this disrupted the process. I'm sorry that you're not able to find the data you want for that test.
Meaning we didn't run the functional tests at all (or, at least, didn't report them).
This run was definitely impacted by dotnet/arcade#10358. I've given them 100 more examples of the problem and notified the issue. You can follow along via the IcM ticket linked.
I run the query again and I only see 4 failures on 8/19 (all RedHat7) no failure afterwards and no failure on main.
This does not look like regression to me @karelz. The test is time sensitive and verifies Cancellation that is best-effort.
My suggestion would be to move to 8 and add more diagnostic and perhaps change
can you please check failure rate @rzikm? It seems like we did not see recent failures and RedHat7 is no longer supported AFAIK. We should probably close it unless there is evidence this is still failing regularly.
Activity
CarnaViire commentedon Aug 18, 2022
Triage: we need to check how frequent this is. We might want to disable the test if it hurts CI. @rzikm could you have a look please?
carlossanlop commentedon Aug 18, 2022
Thanks @CarnaViire . Also please decide if it needs a fix to get backported to RC1, if it meets the bar. If it's a test being disabled, you can backport that too to clean the other PRs CI results.
rzikm commentedon Aug 18, 2022
I have trouble finding this failure in Kusto
returns empty result. Nor does looking for the specific JobId or WorkItemId find anything. @dotnet/dncenghot Can you take a look? It should find at least run from this: https://helixre107v0xdeko0k025g8.blob.core.windows.net/dotnet-runtime-refs-pull-74103-merge-e45453abd21e437ca1/System.Net.Http.Functional.Tests.Attempt.3/2/console.849c5012.log?helixlogtype=result
missymessa commentedon Aug 18, 2022
@rzikm I'm investigating the missing test in the test reporting tables.
missymessa commentedon Aug 18, 2022
@rzikm it looks like these work items ran during the storage outage we had a couple days ago and since we rely on uploading test data to our storage account to ingest into Kusto, this disrupted the process. I'm sorry that you're not able to find the data you want for that test.
ChadNedzlek commentedon Aug 18, 2022
If you look at that PR, this failure is present: System.Net.Http.Functional.Tests Work Item
Meaning we didn't run the functional tests at all (or, at least, didn't report them).
MattGal commentedon Aug 18, 2022
This run was definitely impacted by dotnet/arcade#10358. I've given them 100 more examples of the problem and notified the issue. You can follow along via the IcM ticket linked.
wfurt commentedon Aug 22, 2022
I run the query again and I only see 4 failures on 8/19 (all RedHat7) no failure afterwards and no failure on
main.This does not look like regression to me @karelz. The test is time sensitive and verifies Cancellation that is best-effort.
My suggestion would be to move to 8 and add more diagnostic and perhaps change
runtime/src/libraries/Common/tests/System/Net/Http/HttpClientHandlerTest.Cancellation.cs
Line 269 in 2a2edc8
PassingTestTimeoutkarelz commentedon Aug 23, 2022
Triage: Looks like it was one time hiccup. There does not seem to be more failures. Moving to 8.0.
wfurt commentedon Jun 7, 2023
can you please check failure rate @rzikm? It seems like we did not see recent failures and RedHat7 is no longer supported AFAIK. We should probably close it unless there is evidence this is still failing regularly.
rzikm commentedon Jul 20, 2023
No recent hard failures, only PassedOnRerun, closing now as no action is necessary for time being and RH7 is no longer supported.