Attempt to write outside dump device boundaries

Janos Dohanics web at 3dresearch.com
Fri Jan 14 00:35:45 UTC 2011


I have been having problems with a newly built system which I have
installed 8.2-PRERELEASE amd64. One problem I posted at
http://lists.freebsd.org/pipermail/freebsd-questions/2011-January/225822.html
(Fernando, thank you for replying).

After wiping and reinstalling, I made sure that I could obtain a kernel
dump.

Next, I have set up gmirror. Upon reboot, synchronizing the mirror was
going very slow.

I checked the drives with smartctl; one drive turned out to be faulty.

Next, removed the faulty drive from gmirror, and did reboot -d. At this
point, I got the error "Attempt to write outside dump device
boundaries" similar to what's described in
http://lists.freebsd.org/pipermail/freebsd-stable/2008-December/046982.html

Once the system came back up, and I could verify that the faulty drive
was no longer part of gmirror, did reboot -d again, which caused the
same error.

Next, physically disconnected the faulty drive, and had no problem
getting a kernel dump.

Next, reconnected the faulty drive, but did not insert it into gmirror,
and had no problem getting a kernel dump.

I obviously need to replace the faulty drive. Beyond that, I'm
wondering if someone could explain why is it not possible to get a
kernel dump when a drive in gmirror fails.

-- 
Janos Dohanics
<web at 3dresearch.com>


More information about the freebsd-questions mailing list