New issue
Advanced search Search tips

Issue 910267 link

Starred by 1 user
Status: Fixed
Owner:
mef@chromium.org
Closed: Dec 4
Cc:
fdegans@chromium.org
pauljensen@chromium.org
Components:
EstimatedDays: ----
NextAction: ----
OS: Fuchsia
Pri: 1
Type: Bug-Regression



Sign in to add a comment

[cronet] UrlRequestTest.PerfTest flaked under Fuchsia

Project Member Reported by w...@chromium.org, Nov 29 
UrlRequestTest.PerfTest/0 flaked in https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Fuchsia%20x64/24562 on the Chromium waterfall, with the test timing out:

...
[85/96] BidirectionalStreamDelayRequestHeadersUntilFlush/BidirectionalStreamTest.TestDelayedFlush/0 (113 ms)
[86/96] BidirectionalStreamDelayRequestHeadersUntilFlush/BidirectionalStreamTest.TestDelayedFlush/1 (75 ms)
[00047.157] netstack: loopStreamRead got endpoint error: connection reset by peer (TODO)
Still waiting for the following processes to finish:
	/pkg/cronet_tests__exec --gtest_flagfile=/tmp/.org.chromium.Chromium.oJeabL/.org.chromium.Chromium.bGfdmK --single-process-tests --test-launcher-jobs=4 --test-launcher-output=/tmp/.org.chromium.Chromium.DbmJPk/test_results.xml --test-launcher-retry-limit=0 --test-launcher-summary-output=/data/test_summary.json
... (many repeats) ...
Still waiting for the following processes to finish:
	/pkg/cronet_tests__exec --gtest_flagfile=/tmp/.org.chromium.Chromium.oJeabL/.org.chromium.Chromium.bGfdmK --single-process-tests --test-launcher-jobs=4 --test-launcher-output=/tmp/.org.chromium.Chromium.DbmJPk/test_results.xml --test-launcher-retry-limit=0 --test-launcher-summary-output=/data/test_summary.json
[87/96] UrlRequestTest.TestCancel/0 (1896 ms)
[88/96] UrlRequestTest.TestCancel/1 (722 ms)
[ RUN      ] UrlRequestTest.PerfTest/0
[89/96] UrlRequestTest.PerfTest/0 (TIMED OUT)
...

Not sure whether this might be a Fuchsia netstack regression, or an issue in the test, or in Cronet.

Comment 1 by w...@chromium.org, Nov 29

Cc: fdegans@chromium.org
fdegans: IIUC the test must be hanging at https://chromium.googlesource.com/chromium/src/+blame/HEAD/components/cronet/native/test/url_request_test.cc#932 because the connection got reset; would that fit with the not-readable-on-EOF issue that you were seeing in the HTTP-service tests?

Comment 2 by fdegans@chromium.org, Nov 29 

Yes that sounds very similar. I'll see if I can repro and confirm that's the same issue.

Comment 3 by mef@chromium.org, Dec 4 

I don't see cronet_tests failures on https://ci.chromium.org/p/chromium/builders/luci.chromium.ci/Fuchsia%20x64?limit=500 since 2018-11-30, does it mean that "not-readable-on-EOF issue" was the culprit and is now fixed?

Comment 4 by fdegans@chromium.org, Dec 4

Status: Fixed (was: Untriaged)
There was a fix for that issue in Fuchsia but it didn't cover all cases as we are still seeing failures. It could have fixed that specific one though. Let's close this for now. We can reopen if it happens again.
►

Sign in to add a comment