test failure: os-autoinst: 29-backend-driver.t:31 Failed test 'exit logged'
All 15.3 builds are failing now with the same error
[ 211s] 3: [17:11:16] ./28-signalblocker.t ..................... ok 686 ms ( 0.00 usr 0.00 sys + 0.56 cusr 0.14 csys = 0.70 CPU) [ 212s] 3: [ 212s] 3: # Failed test 'exit logged' [ 212s] 3: # at ./29-backend-driver.t line 31. [ 212s] 3: # STDOUT & STDERR: [ 212s] 3: # [37m[2021-12-06T17:11:16.953486Z] [debug] received magic close [ 212s] 3: # [0m [ 212s] 3: # don't match: [ 212s] 3: # (?^u:backend.*exited) [ 212s] 3: # as expected [ 212s] 3: [37m[2021-12-06T17:11:17.057808Z] [debug] backend process exited: 72057594037927935 [ 212s] 3: [0m# Looks like you failed 1 test of 12. [ 212s] 3: [17:11:17] ./29-backend-driver.t .................... [ 212s] 3: Dubious, test returned 1 (wstat 256, 0x100) [ 212s] 3: Failed 1/12 subtests
Also the output of 10-virtio_terminal.t looks weird, but not sure if this was happening before.
It still succeeds, though:
[ 196s] 3: Can't open out pipe for reading Interrupted system call at ./10-virtio_terminal.t line 44. [ 196s] 3: [ 196s] 3: # Failed test 'no (unexpected) warnings (via END block)' [ 196s] 3: # at /usr/lib/perl5/5.26.1/Test/Builder.pm line 135. [ 196s] 3: # Got the following unexpected warnings: [ 196s] 3: # 1: Can't open out pipe for reading Interrupted system call at ./10-virtio_terminal.t line 44. [ 196s] 3: [33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[17:11:01] ./10-virtio_terminal.t ................... ok 820 ms ( 0.01 usr 0.00 sys + 0.71 cusr 0.13 csys = 0.85 CPU)
This is how it looks like on 15.2:
[ 164s] 3: [33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[33m[0m[10:16:34] ./10-virtio_terminal.t ................... ok 835 ms ( 0.01 usr 0.00 sys + 0.71 cusr 0.14 csys = 0.86 CPU)
In any case we should get rid of the ansi codes.
Created an issue here: https://github.com/mudler/Mojo-IOLoop-ReadWriteProcess/issues/29
#4 Updated by kraih about 2 months ago
Bisected the problem, and the responsible change is https://github.com/mudler/Mojo-IOLoop-ReadWriteProcess/commit/1e0addb6cee6ffd74e9e611270fa9eb8759a6976.
#6 Updated by mkittler about 1 month ago
- Status changed from New to Feedback
#7 Updated by mkittler about 1 month ago
I suppose we're now focusing on https://github.com/mudler/Mojo-IOLoop-ReadWriteProcess/pull/33.
I've already tested it against the os-autoinst test and it works. I've also conducted a test run (with no browser tab open) and it passed normally.
In the meantime I also adjusted our devel repos. TW user should lock the RWP package for now (
zypper al perl-Mojo-IOLoop-ReadWriteProcess).
#8 Updated by kraih about 1 month ago
Quick way to verify that PR#33 resolves the issue:
$ sudo zypper in perl-App-cpanminus ... $ cd os-autoinst $ cpanm -n -L extlib https://api.github.com/repos/mudler/Mojo-IOLoop-ReadWriteProcess/tarball/d816ead3959fb876d632654bdb1ee7fd47c5633d ... $ prove -I. -Iextlib/lib/perl5/ t/29-backend-driver.t t/29-backend-driver.t .. ok All tests successful. Files=1, Tests=12, 0 wallclock secs ( 0.02 usr 0.01 sys + 0.12 cusr 0.06 csys = 0.21 CPU) Result: PASS
#11 Updated by mkittler about 1 month ago
The PR has been merged and it is part of the tag 0.32 which should make its way to Factory/TW automatically (see https://build.opensuse.org/request/show/939161).
When it is in Factory, I'll restore our linking in the devel projects. The new version isn't strictly required but having the race condition fixed (the one which Clemens wanted to fix initially) is likely a good idea.