Lustre discuss - Apr 2010 - Kernel Panic on osc:osc_statfs_interpret

Hey everyone,

I have a machine that is constantly crashing with a Kernel Panic at
osc:osc_statfs_interpret.
The machine had almost no Lustre activity going on at the time of the
crash. When it came up I ran some ls and dd tests at first to verify
that it was indeed working.
I don''t have any other nodes exhibiting the same symptoms so
I''m
leaning towards this being a hardware issue on the problem node.

Any insight into the osc:osc_statfs_interpret function and what it
might touch that could cause a KP?

Sorry for attaching a picture of it but it''s a Dell and the remote
console software won''t let me copy and paste.

Erik
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screen shot 2010-04-28 at 4.10.05 PM.png
Type: image/png
Size: 39572 bytes
Desc: not available
Url :
http://lists.lustre.org/pipermail/lustre-discuss/attachments/20100429/c39cc661/attachment-0001.png

Hi Erik,

It looks like:
https://bugzilla.lustre.org/show_bug.cgi?id=20482

It''s a problem with how it handles inactive OSTs. It''s fixed
it 1.8.2,
but the work around seems to be to do --writeconf everywhere to remove 
the OSTs from the config.

- Kit

On 4/29/2010 11:31 AM, Erik Froese wrote:
> Hey everyone,
>
> I have a machine that is constantly crashing with a Kernel Panic at
> osc:osc_statfs_interpret.
> The machine had almost no Lustre activity going on at the time of the
> crash. When it came up I ran some ls and dd tests at first to verify
> that it was indeed working.
> I don''t have any other nodes exhibiting the same symptoms so
I''m
> leaning towards this being a hardware issue on the problem node.
>
> Any insight into the osc:osc_statfs_interpret function and what it
> might touch that could cause a KP?
>
> Sorry for attaching a picture of it but it''s a Dell and the remote
> console software won''t let me copy and paste.
>
> Erik
>    

-- 
---
Kit Westneat
kwestneat at datadirectnet.com
812-484-8485

Thanks Kit,

We found that bug yesterday and it is indeed what bit us.
The bugreport says the fix landed for 1.8.1.1. Our servers were
running 1.8.1.1 and the clients were on 1.8.1.

We''ve tested a few clients with the 1.8.1.1 version and it works just
fine so I''m in the process of rolling out the upgraded rpms.

Thanks for helping out!

Erik


On Thu, Apr 29, 2010 at 4:10 PM, Kit Westneat <kwestneat at ddn.com>
wrote:
> Hi Erik,
>
> It looks like:
> https://bugzilla.lustre.org/show_bug.cgi?id=20482
>
> It''s a problem with how it handles inactive OSTs. It''s
fixed it 1.8.2, but
> the work around seems to be to do --writeconf everywhere to remove the OSTs
> from the config.
>
> - Kit
>
> On 4/29/2010 11:31 AM, Erik Froese wrote:
>>
>> Hey everyone,
>>
>> I have a machine that is constantly crashing with a Kernel Panic at
>> osc:osc_statfs_interpret.
>> The machine had almost no Lustre activity going on at the time of the
>> crash. When it came up I ran some ls and dd tests at first to verify
>> that it was indeed working.
>> I don''t have any other nodes exhibiting the same symptoms so
I''m
>> leaning towards this being a hardware issue on the problem node.
>>
>> Any insight into the osc:osc_statfs_interpret function and what it
>> might touch that could cause a KP?
>>
>> Sorry for attaching a picture of it but it''s a Dell and the
remote
>> console software won''t let me copy and paste.
>>
>> Erik
>>
>
>
> --
> ---
> Kit Westneat
> kwestneat at datadirectnet.com
> 812-484-8485
>
>