[Bug 1915126] Re: autopkgtest times out (or fails before that) in hirsute on test infra, passes locally

Fri Feb 12 16:32:04 UTC 2021

Ok, here you go. I reproduced the (or a) failure in Canonistack. This
was the first run FWIW, so it wasn't too hard to get it to fail:

autopkgtest [14:05:12]: test systemd-fsckd: [-----------------------
test_fsck_failure (__main__.FsckdTest)
Ensure that a failing fsck doesn't prevent fsckd to stop ... bash: line 1:  1250 Killed                  /tmp/autopkgtest.5SlVJo/build.mmt/src/debian/tests/systemd-fsckd 2> >(tee -a /tmp/autopkgtest.5SlVJo/systemd-fsckd-stderr >&2) > >(tee -a /tmp/autopkgtest.5SlVJo/systemd-fsckd-stdout)
autopkgtest [14:05:16]: test process requested reboot with marker test_fsck_failure
autopkgtest-virt-ssh: WARNING: ssh connection failed. Retrying in 3 seconds...
test_fsck_failure (__main__.FsckdTest)
Ensure that a failing fsck doesn't prevent fsckd to stop ... FAIL

======================================================================
FAIL: test_fsck_failure (__main__.FsckdTest)
Ensure that a failing fsck doesn't prevent fsckd to stop
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/tmp/autopkgtest.5SlVJo/build.mmt/src/debian/tests/systemd-fsckd", line 190, in test_fsck_failure
    self.check_plymouth_start()
  File "/tmp/autopkgtest.5SlVJo/build.mmt/src/debian/tests/systemd-fsckd", line 158, in check_plymouth_start
    self.assertUnitStarted(unit)
  File "/tmp/autopkgtest.5SlVJo/build.mmt/src/debian/tests/systemd-fsckd", line 228, in assertUnitStarted
    self.assertTrue(has_unit_started(unit))
AssertionError: False is not true

----------------------------------------------------------------------
Ran 1 test in 11.289s

FAILED (failures=1)
autopkgtest [14:06:11]: test systemd-fsckd: -----------------------]
autopkgtest [14:06:12]: test systemd-fsckd:  - - - - - - - - - - results - - - - - - - - - -
systemd-fsckd        FAIL non-zero exit status 1
autopkgtest [14:06:13]: @@@@@@@@@@@@@@@@@@@@ summary
timedated            PASS
hostnamed            PASS
localed-locale       PASS
localed-x11-keymap   PASS
logind               PASS
unit-config          PASS
storage              PASS
networkd-test.py     PASS
build-login          PASS
boot-and-services    PASS
udev                 PASS
root-unittests       PASS
tests-in-lxd         PASS
upstream             PASS
boot-smoke           PASS
systemd-fsckd        FAIL non-zero exit status 1

Full logs attached. The commandline is at the top of the log - is that
enough for you to be able to have a go, Balint? I suggest adding
--shell-fail of course so you can break in once it happens.

** Attachment added: "systemd-canonistack-cpu4-ram8-disk50.tar.xz"
   https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1915126/+attachment/5463223/+files/systemd-canonistack-cpu4-ram8-disk50.tar.xz

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to systemd in Ubuntu.
https://bugs.launchpad.net/bugs/1915126

Title:
  autopkgtest times out (or fails before that) in hirsute on test infra,
  passes locally

Status in Auto Package Testing:
  Incomplete
Status in systemd package in Ubuntu:
  Invalid

Bug description:
  Hi,
  I've asked yesterday on IRC but so far got no answer. I think it is right to file a bug about the current state of systemd autopkgtest to unite the efforts in regard to it.

  I was looking at the systemd tests for a no-change rebuild that really
  had no reason to now make it fail. While checking I found that as of
  this month (first bad test on 1st of February) most of the systemd
  test runs on amd64 will not pass/fail but instead in most cases time
  out.

  I've seen various people retry the case as it shows the typical
  symptoms of a "not the fault of the package, let us retry this" case.
  But it seems that won't help as the test history is rather clear.

  * Watch this in monospace to make more sense of it *

  $ check-autopkgtest-stats.sh -c 50 -p systemd -r "hirsute" -a "amd64"
  Check last 50 test results for src:systemd on releases 'hirsute' on architectures 'amd64'
  Of the 50 last tests, we had these subtest failing per release/arch:

  hirsute
    amd64
      tests-in-lxd                   (F  2% f  0% S  0% B 12% => P 52%/) .BTTTBTT.BTTTTTBTTTBBTTT..F......................
      hostnamed                      (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      build-login                    (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      unit-config                    (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      networkd-testpy                (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      localed-locale                 (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      boot-and-services              (F  4% f  0% S  0% B 12% => P 50%/) .BTTTBTT.BTTTTTBTTTBBTTT...............F.....F...
      timedated                      (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      boot-smoke                     (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      logind                         (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      storage                        (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      upstream                       (F 12% f  0% S  0% B 12% => P 42%/) .BTTTBTT.BTTTTTBTTTBBTTTF.....F.F..F........F..F.
      udev                           (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      systemd-fsckd                  (F  8% f  0% S  0% B 12% => P 46%/) FBTTTBTTFBTTTTTBTTTBBTTT.............F......F....
      root-unittests                 (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................
      localed-x11-keymap             (F  0% f  0% S  0% B 12% => P 54%/) .BTTTBTT.BTTTTTBTTTBBTTT.........................

  We see that formerly we had the known to be somewhat flaky, but
  otherwise working test (time goes from right to left). So we had the
  usual suspects of tests like upstream/systemd-fsck that failed a few
  times, but also working runs in between.

  But since February 22/24 runs failed very bad.
  6 (=B) of those cases are aborting mid execution
    autopkgtest [22:08:58]: ERROR: testbed failure: testbed auxverb failed with exit code 255
  And 16 of them timed out
    <VirtSubproc>: failure: Timed out on waiting for ssh connection

  Sadly the 2/24 that didn't hard fail by that where broken by the known
  flaky systemd-fsckd test.

  Something drives this test crazy that we have to find and resolve, at
  the current rate nothing depending on it is likely to migrate.

To manage notifications about this bug go to:
https://bugs.launchpad.net/auto-package-testing/+bug/1915126/+subscriptions