[Bug 232671] [gmirror] gmirror fails to recover from degraded mirror sets in some circumstances

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=232671

--- Comment #2 from Mark Johnston <markj at FreeBSD.org> ---
Nice.  JFYI, there's a number of test cases in tests/sys/geom/class/mirror
which attempt to reproduce issues like the one you described.

Regarding the proposed fix, check out r306743.  It aimed to fix a
similar-sounding problem.  I think it's not quite sufficient in that we make
the decision to transition to RUNNING before checking for broken disks.  In the
scenario you described, it sounds like we should transition back to STARTING
and wait for a third disk to arrive.

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