https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=288345

--- Comment #41 from Craig Leres <[email protected]> ---
(In reply to Craig Leres from comment #40)
The GENERIC-DEBUG finally produced a panic/crashdump. About 15 minutes into the
scheduled poudriere bulk build this morning it panic'ed. The ipmi console
showed:

    panic: not reclaimed nullfs vnode

and a kgdb prompt. It clearly had saved the crash dump. Naively I entered
"continue" and was relegated to waiting for it to save the crash dump again...
Then I entered "reset" and the system rebooted, and saved the crash dump which
is now here:

    https://ee.lbl.gov/.bsd-C6sfr/

I think it's interesting that D state problems most commonly occur during the
first ~30 minutes of the build. I know the package build rate is high at first
as many quick-to-build ports are built. I'm guessing a test script that
repeatedly creates and deletes zfs partitions in the way that poudriere does
would trigger this bug.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to