On 220131 0806, Thomas Huth wrote: > > On 29/1/22 14:34, Peter Maydell wrote: > > > Hi; the build-oss-fuzz gitlab CI job seems to intermittently > > > but quite commonly hit the 1 hour timeout mark and get killed. > > > Examples from the last couple of days: > > > > > > https://gitlab.com/qemu-project/qemu/-/jobs/2030815488 > > > https://gitlab.com/qemu-project/qemu/-/jobs/2029246068 > > > https://gitlab.com/qemu-project/qemu/-/jobs/2029013479 > > > https://gitlab.com/qemu-project/qemu/-/jobs/2024871970 > > > https://gitlab.com/qemu-project/qemu/-/jobs/2022584981 > > > > > > Can we do anything to cut down on the runtime of this job > > > and make it reliably complete inside the time limit? > > All the jobs that you've listed hang in the very same test > (qtest-i386/boot-serial-test), so I assume it's rather a test that now hangs > occasionally, and not a generic slowness (otherwise the jobs would fail > sometimes earlier, sometimes later). >
Judging by this job, from one of those pipelines, the big qos-test is next up after boot-serial-test (the last test that finishes successfully): https://gitlab.com/qemu-project/qemu/-/jobs/2024872156#L57 So something is likely timing out there, when built with ASAN.. -Alex > Thus we likely have a regression in the code that only shows up occasionally > in these builds... Can you mark a point in time when these issues first > happened? > > Thomas >
