[PC-BSD Testing] Panic from ZFS

Bill Leeper protagonist at charter.net
Fri Oct 24 08:53:13 PDT 2008


I was wondering if you would speak up again. Also, for those  
interested here is the link you gave us back in the alpha testing.

  http://kerneltrap.org/FreeBSD/ZFS_Stability

I have to admit I am one of those running on a less than optimal  
system, having a 32 bit processor and only 2gb of RAM, but so far I  
have had no problems at all with it. But then this is running on a  
system with no critical data so if things go wrong it is no big deal.



On Oct 24, 2008, at 8:07 AM, Edward O'Callaghan wrote:

> Actually Kris,
>
> My point was that you could suffer complete data corruption !
> I know about the message as I asked for it to be put there,  
> remember ;)
>
> I really do think it should be disabled on systems where it should not
> be running on until you are using FreeBSD 8 with the update to version
> 11. That is, that I can now confirm that it does indeed resolve the
> memory issue.
>
> Regards,
> Edward.
>
> 2008/10/25 Kris Moore <kris at pcbsd.com>:
>> Arthur wrote:
>>> At 02:49 AM 10/24/2008, you wrote:
>>>> On Fri, Oct 24, 2008 at 7:30 AM, Edward O'Callaghan
>>>> <victoredwardocallaghan at gmail.com> wrote:
>>>>> Hi,
>>>>>
>>>>> You can get version 6 of zfs to panic in FreeBSD by doing the  
>>>>> following;
>>>>> * Have 512 or less MB of RAM
>>>>> * Fill most of your zpool up then run a zpool scrub on the pool..
>>>>>
>>>>> This is a out-of-memory kernel panic that particularly affects  
>>>>> i386.
>>>>>
>>>>> My suggested work around is to put the following into loader.conf
>>>>> vfs.zfs.zil_disable=1
>>>>> vfs.zfs.prefetch_disable=1
>>>>>
>>>>> N.B. Disabling ZIL can have serious consequences depending on
>>>> your workload.
>>>>> FreeBSD 8 upgrades ZFS to version 11. However I am not sure how  
>>>>> this
>>>>> affects things as this only happens on FreeBSD and not Solaris.
>>>>>
>>>>> Kris, As I said before. I highly recommend you disable ZFS  
>>>>> support on
>>>>> 32bit systems with less then 1gb of ram else people _could_ lose  
>>>>> data
>>>>> ?
>>>> I've always believed that there is a HEADS UP somewhere that said
>>>> something about dedicating at least 1GB for ZFS on FreeBSD.
>>>
>>> Yes, I remember this as well.
>>>
>>> Arthur
>>
>>
>> Correct! If you try to install with ZFS, this message appears  
>> first, and
>> you have to click "OK" to continue:
>>
>> "Warning: ZFS usage is primarily for high-end systems with a 64 bit
>> processor and 4+ GB of memory. It may run on lesser hardware, but  
>> is not
>> recommended."
>>
>> That pretty much covers all the bases here. I don't want to outright
>> disable ZFS, since some users have enjoyed testing / playing with  
>> it on
>> systems with 512MB of ram in qemu / vmware or on a test desktop. Its
>> more of a learning experience for them to just play with zfs and  
>> see how
>> it works. (I've done that myself a bit, on some of my older systems)
>> However, again, the message does let users know that it should have  
>> 4GB
>> of memory, and use the amd64 release if possible.
>>
>>
>> --
>>
>> Kris Moore
>> PC-BSD Software
>> http://www.pcbsd.com
>> _______________________________________________
>> Testing mailing list
>> Testing at lists.pcbsd.org
>> http://lists.pcbsd.org/mailman/listinfo/testing
>>
>
>
>
> -- 
> All Documents adhered to the ISO/IEC 26300  standard file format for
> electronic office documents, such as spreadsheets, charts,
> presentations and word processing documents from this email address.
> The author does not take responsibility of the recipients inability to
> read international standards and who use proprietary products such as
> MS Office.
> See: http://www.openoffice.org/
> Edward O'Callaghan.
> --
> http://fedoraproject.org/wiki/EdwardOcallaghan
> http://moonshine.opn4.org/
> http://www.pcbsd.org/
> _______________________________________________
> Testing mailing list
> Testing at lists.pcbsd.org
> http://lists.pcbsd.org/mailman/listinfo/testing


Computers can be fun again
                 format C:\
   Alternatives are out there




More information about the Testing mailing list