Re: [PATCH] tests: Disable migration-test

["Dr. David Alan Gilbert" <dgilbert@redhat.com>]

* Peter Maydell (peter.maydell@linaro.org) wrote:
> The migration-test is annoyingly flaky. Examples:
> 
> https://gitlab.com/qemu-project/qemu/-/jobs/3806090216
> (a FreeBSD job)
>   32/648 ERROR:../tests/qtest/migration-helpers.c:205:wait_for_migration_status: assertion failed: (g_test_timer_elapsed() < MIGRATION_STATUS_WAIT_TIMEOUT) ERROR
> 
> on a local macos x86 box:
> ▶  34/621 ERROR:../../tests/qtest/migration-helpers.c:151:migrate_query_not_failed: assertion failed: (!g_
> str_equal(status, "failed")) ERROR
>  34/621 qemu:qtest+qtest-i386 / qtest-i386/migration-test                         ERROR          168.12s   killed by signal 6 SIGABRT
> ――――――――――――――――――――――――――――――――――――― ✀  ―――――――――――――――――――――――――――――――――――――
> stderr:
> qemu-system-i386: Failed to peek at channel
> query-migrate shows failed migration: Unable to write to socket: Broken pipe
> **
> ERROR:../../tests/qtest/migration-helpers.c:151:migrate_query_not_failed: assertion failed: (!g_str_equal(status, "failed"))
> 
> (test program exited with status code -6)
> ――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
> 
> ▶  37/621 ERROR:../../tests/qtest/migration-helpers.c:151:migrate_query_not_failed: assertion failed: (!g_str_equal(status, "failed")) ERROR
>  37/621 qemu:qtest+qtest-x86_64 / qtest-x86_64/migration-test                     ERROR          174.37s   killed by signal 6 SIGABRT
> ――――――――――――――――――――――――――――――――――――― ✀  ―――――――――――――――――――――――――――――――――――――
> stderr:
> query-migrate shows failed migration: Unable to write to socket: Broken pipe
> **
> ERROR:../../tests/qtest/migration-helpers.c:151:migrate_query_not_failed: assertion failed: (!g_str_equal(status, "failed"))
> 
> (test program exited with status code -6)
> ――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
> 
> I've seen this on other CI jobs as well, but Gitlab's UI makes it
> pretty much impossible to re-find failed jobs, since you can't
> search for them by failure reason at all.
> 
> I've also seen this fail on the OpenBSD vm build.
> 
> I've seen the migration-test hang on the s390 private CI runner
> in such a way that even though the CI job has timed out, the
> stale QEMU and migration-test processes are still lying around on
> the host.
> 
> I've complained about these before, but nobody has either investigated
> or suggested improvements to the test program that would let us gather
> more information about what's happening when these fail.
>  https://lore.kernel.org/qemu-devel/CAFEAcA8x_iM3hN2-P9F+huXnXFXy+D6FzE+Leq4erLdg7zkVGw@mail.gmail.com/

Damn this is really going to impact the stability of migration if we
don't regularly test.
But fundamentally, I've never been able to debug much of the reports
that come from flakyness in gitlab ci; we're not getting the most basic
information like which subtest or where we're upto in the test which
makes it very very hard to debug.

Dave

> So this is the big hammer: disable the test entirely, so that we
> don't keep getting CI job intermittent failures because of it.
> When somebody has time to investigate, we can fix the underlying
> cause and reenable the job.
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> This is an "if you don't want this, propose something else" patch :-)
> ---
>  tests/qtest/meson.build | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/tests/qtest/meson.build b/tests/qtest/meson.build
> index 29a4efb4c24..0e362fcb1e0 100644
> --- a/tests/qtest/meson.build
> +++ b/tests/qtest/meson.build
> @@ -11,6 +11,12 @@ slow_qtests = {
>    'test-hmp' : 120,
>  }
>  
> +# Temporarily disabled tests can be listed here
> +qtests_disabled = [
> +  # This test is currently flaky and fails intermittently
> +  'migration-test',
> +]
> +
>  qtests_generic = [
>    'cdrom-test',
>    'device-introspect-test',
> @@ -343,6 +349,9 @@ foreach dir : target_dirs
>    endif
>  
>    foreach test : target_qtests
> +    if test in qtests_disabled
> +      continue
> +    endif
>      # Executables are shared across targets, declare them only the first time we
>      # encounter them
>      if not qtest_executables.has_key(test)
> -- 
> 2.34.1
> 
-- 
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK