Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 27 Aug 2009 13:18:06 +0400
From:      "Mikhail (Plus Plus)" <m@plus-plus.su>
To:        freebsd-fs@freebsd.org
Subject:   Re: need help with ZFS
Message-ID:  <4A964F4E.4080009@plus-plus.su>
In-Reply-To: <4A927CB3.3040402@plus-plus.su>
References:  <4A927CB3.3040402@plus-plus.su>

next in thread | previous in thread | raw e-mail | index | archive | help
Last night I installed FreeBSD-8.0BETA3 AMD64, upgraded zpool and zfs, 
and this morning I did some load on FS to see if it runs more stable now.

To stress-load the system I ran rsync from ZFS to mounted UFS volume and 
at the same time I started bonie++ benchmark on ZFS volume:

bonnie++ -s17000 -d. -n64

Everything worked fine for about ~90 minutes, but then system paniced. I 
got escaped to "db> " prompt on local console, but I have zero debugging 
experience, so I just rebooted the server. The only thing I noticed was 
that last mentioned PID on the panic screen was bonnie++'s process.

Also, I'm very confused after re-reading ZFS tuning wiki from here:
http://wiki.freebsd.org/ZFSTuningGuide

"amd64
FreeBSD 7.2+ has improved kernel memory allocation strategy and no 
tuning may be necessary on systems with more than 2 GB of RAM."

Does that mean I no longer have to tune ZFS via loader.conf? Just better 
leave it empty on FreeBSD-8.0 installation?

Right now I'm going to continue with rsync and will start bonnie++ in 
parallel keeping my loader.conf with the following values:

vm.kmem_size="1536M"
vm.kmem_size_max="3072M"
vm.pmap.shpgperproc="1024"
vfs.zfs.arc_min="256M"
vfs.zfs.arc_max="384M"
vfs.zfs.vdev.cache.size="50M"
vfs.zfs.prefetch_disable="1"
kern.maxproc="20000"

I would appreciate any tips on showing me how to debug system panic from 
"db> " prompt in case system panics again..

Thank you.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4A964F4E.4080009>