upgrade to 10 for zfs ?

[Fabian Keil]

Albert Shih <Albert.Shih at obspm.fr> wrote:

> I've (in production) a file server running FreeBSD 9.1-p8. The only thing
> this server does is file-server. Not event through nfs just keep all
> file.
> 
> All data file system is using ZFS. I got pretty big (or let's say not
> small) pool (150To) and the server got 48Go of Ram. 
> 
> Since we using zfs send/received from a another server I got daily a zfs
> core dump. I failed to find which process make this core-dump but it's
> definitvly zfs : 
> 
> pid 23943 (zfs), uid 0: exited on signal 6 (core dumped)
> pid 70462 (zfs), uid 0: exited on signal 6 (core dumped)
> pid 71042 (zfs), uid 0: exited on signal 6 (core dumped)
> 
> I check every zfs file system and don't find any corup data. 
> 
> One of the zfs sender is running FreeBSD 10 the other linux/Debian. 
> 
> What's you advise ? Should I upgrade to FreeBSD 10. ? 

Note that receiving zfs processes may also crash on FreeBSD 11:
http://lists.freebsd.org/pipermail/freebsd-fs/2014-July/019814.html

The cause hasn't been analysed yet and from your description
it's not clear to me if it's the same assertion that's being
triggered, but in my case the problem is reproducible, so you
could look for receiving datasets that are missing snapshots
that should have been received already.

If no snapshots are missing, it's probably an unrelated issue.

Obviously checking the backtrace wouldn't hurt either.

Fabian
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/freebsd-questions/attachments/20140807/3b2082db/attachment.sig>