ZFS Crash

[Larry Rosenman]

On Sat, 30 May 2009, Larry Rosenman wrote:

> On Fri, 29 May 2009, Kip Macy wrote:
>
>>> Is there anything else I can supply?  Is the minute-by-minute
>>> vmstat info useful?  Should I keep running it?
>> 
>> Not until I can test out changes that bound ZFS memory consumption
>> when doing compression.
>> 
>>> I'm willing to be a guinea pig for code changes.
>> 
>> Cool. I'll send you a patch when I get something.
> I had a 2nd crash.  I've added the textdump and crashstats files to
> http://www.lerctr.org/~ler/ZFS_CRASH
>
> This was also with a full backup.   The backtrace that was on the screen is 
> in the textdump
> so I didn't bother to get the picture.
And a 3rd crash.  I had set debug.ddb.textdump.pending to 0 hoping that
it would cause the scripts to run.  It didn't, and we didn't get
a textdump at all (system rebooted on it's own).

Is there a limit I can set that would leave enough memory for the
(de)compression stuff should have enough memory?

Currently running with no tunables set. 
Loader.conf:
-----
#geom_raid3_load="YES"
#boot_verbose="-v"
kern.ipc.semmns=4096
kern.ipc.semmni=2048
machdep.hyperthreading_allowed=1
zfs_load="YES"
cpuctl_load="YES"
###Temp till increased by RE@###
#vm.kmem_size_max=1G
#vm.kmem_size_max=4294967296
#vm.kmem_size_max=4g
#vfs.zfs.arc_max=4g
###Temp till increased by RE@###
#kern.hwpmc.mtxpoolsize=256
#kern.hwpmc.nsamples=256
#kern.hwpmc.hashsize=256
#kern.hwpmc.callchaindepth=8
#kern.hwpmc.nbuffers=256
#kern.hwpmc.logbuffersize=4
----


-- 
Larry Rosenman                     http://www.lerctr.org/~ler
Phone: +1 512-248-2683                 E-Mail: ler at lerctr.org
US Mail: 430 Valona Loop, Round Rock, TX 78681-3893