On Mon, May 16, 2022 at 09:08:45PM +0200, Paul Gevers wrote: > Hi Julien, > > On 16-05-2022 14:10, Julien Cristau wrote: > > Control: severity -1 normal > > > > On Fri, Apr 15, 2022 at 10:36:54PM +0200, Paul Gevers wrote: > > > I looked at the results of the autopkgtest of you package because it was > > > showing up as a regression for the upload of python3-defaults. I noticed > > > that the test regularly fails on several architectures. Most peculiarly on > > > amd64, the latest version seems to pass on ci-worker13 (our most powerful > > > host looking at number of cores, memory and disk space) and fails > > > (timeout) > > > on the other amd64 hosts.
What are the specs of these hosts? How long are tests allowed to take? > > > Because the unstable-to-testing migration software now blocks on > > > regressions in testing, flaky tests, i.e. tests that flip between > > > passing and failing without changes to the list of installed packages, > > > are causing people unrelated to your package to spend time on these > > > tests. > > > > > Which specific tests are you having issues with? > > It seems not at all specific: > > E.g. > https://ci.debian.net/data/autopkgtest/testing/amd64/m/mercurial/21333636/log.gz > ends with: > test-encode.t > test-encode.t ... # Test test-encode.t > # Running sh "/tmp/hgtests.qsr01vxk/child797/test-encode.t.sh" > # Timout reached for process 98896 > > https://ci.debian.net/data/autopkgtest/testing/amd64/m/mercurial/21329273/log.gz > ends with: > test-share-bookmarks.t#vfs#normal > test-share-bookmarks.t#vfs#normal ... # Test > test-share-bookmarks.t#vfs#normal > # Timout reached for process 77193 > # Running sh > "/tmp/hgtests.f1fx8cv7/child370/test-share-bookmarks.t-vfs-normal.sh" > > https://ci.debian.net/data/autopkgtest/testing/amd64/m/mercurial/20890959/log.gz > ends with: > test-debugbackupbundle.t > test-debugbackupbundle.t ... # Test test-debugbackupbundle.t > # Running sh "/tmp/hgtests.d5oe121c/child855/test-debugbackupbundle.t.sh" > # Timout reached for process 100461 > > https://ci.debian.net/data/autopkgtest/testing/amd64/m/mercurial/20858105/log.gz > ends with: > test-status.t#dirstate-v1 > test-status.t#dirstate-v1 ... # Test test-status.t#dirstate-v1 > # Timout reached for process 56796 > # Running sh "/tmp/hgtests.p1egbody/child227/test-status.t-dirstate-v1.sh" > AFAICT the above are expected messages, and not errors. When autopkgtest itself errors out with "ERROR: timed out on command ..." it would probably be useful if it showed which processes were running at the time it gave up so it's maybe a bit easier to tell what got stuck. > https://ci.debian.net/data/autopkgtest/testing/amd64/m/mercurial/20816570/log.gz > ends with: > Failed test-hghave.t: output changed > Failed test-hgrc.t: output changed > Failed test-https.t: output changed > Failed test-parseindex.t: output changed > Failed test-patchbomb-tls.t: output changed > Failed test-paths.t: output changed > Failed test-run-tests.t: output changed > # Ran 918 tests, 47 skipped, 7 failed. > python hash seed: 1267217074 > # Timout reached for process 101372 > # Cleaning up HGTMP /tmp/hgtests.fo156zei > make: *** [Makefile:140: tests] Error 1 > > https://ci.debian.net/data/autopkgtest/testing/arm64/m/mercurial/21443035/log.gz > ends with > Failed test-convert-bzr.t: output changed > # Ran 918 tests, 47 skipped, 1 failed. > python hash seed: 2382336912 > # Timout reached for process 101344 > # Cleaning up HGTMP /tmp/hgtests.a0d42699 > make: *** [Makefile:140: tests] Error 1 These two are actual test failures, not timeouts. The first set were fixed in 6.1.2-1; I'm not familiar with the test-convert-bzr.t failure, could be a race condition in the test. Either way, it's unrelated to the "timeouts" issue. Cheers, Julien