https://blogs.gentoo.org/ago/2020/07/04/gentoo-tinderbox/ Issue: dev-vcs/mercurial-5.8.1 fails tests on stable ppc. Discovered on: ppc (internal ref: ppc_at)
Created attachment 777779 [details] build.log build log and emerge --info
Created attachment 867013 [details] build.log (ppc, 6.4.4) More test failures compared to 5.8.1 but no regression (less tests in 5.8.1). [...] Failed test-debugcommands.t: output changed Failed test-dirstate-read-race.t#dirstate-v1#pre-all-read: output changed Failed test-dirstate-read-race.t#dirstate-v1#pre-some-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-append#pre-all-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-append#pre-some-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-rewrite#pre-all-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-rewrite#pre-some-read: output changed Failed test-dirstate-status-write-race.t#dirstate-v1: output changed Failed test-dirstate-status-write-race.t#dirstate-v2-append: output changed Failed test-dirstate-status-write-race.t#dirstate-v2-rewrite: output changed Failed test-remotefilelog-gc.t: output changed # Ran 925 tests, 49 skipped, 11 failed. python hash seed: 3320138338 * ERROR: dev-vcs/mercurial-6.4.4::gentoo failed (test phase): * Tests fail with python3.11
Created attachment 867143 [details] build.log (ppc, 6.5) Failed test-clone-stream.t#stream-bundle2-v3: output changed Failed test-debugcommands.t: output changed Failed test-dirstate-read-race.t#dirstate-v1#pre-all-read: output changed Failed test-dirstate-read-race.t#dirstate-v1#pre-some-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-append#pre-all-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-append#pre-some-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-rewrite#pre-all-read: output changed Failed test-dirstate-read-race.t#dirstate-v2-rewrite#pre-some-read: output changed Failed test-dirstate-status-write-race.t#dirstate-v1: output changed Failed test-dirstate-status-write-race.t#dirstate-v2-append: output changed Failed test-dirstate-status-write-race.t#dirstate-v2-rewrite: output changed Failed test-lock-badness.t: output changed # Ran 934 tests, 50 skipped, 12 failed. python hash seed: 2313753079 * ERROR: dev-vcs/mercurial-6.5::gentoo failed (test phase): * Tests fail with python3.11
Those look like timeouts to me...
(In reply to Sam James from comment #4) > Those look like timeouts to me... Looks like we could try setting HGTEST_TIMEOUT?
(In reply to Sam James from comment #5) > (In reply to Sam James from comment #4) > > Those look like timeouts to me... > > Looks like we could try setting HGTEST_TIMEOUT? OK but to which value? I guess it depends on the speed of the testing machine.
I've consistently seen the test-dirstate-read-race.t failures on my mostly-stable hardened-amd64 machines, even with HGTEST_TIMEOUT bumped from 360 to 2400---and while I do routinely have to increase timeouts to address test failures in other packages, usually tripling a timeout is more than enough. So I wonder if this is simply a brittle test (especially since from the name I guess it's supposed to be testing resilience to a particular race condition).
