System hangs during heavy sequential write to mfi device

[Ireneusz Pluta]

Antony Mawer pisze:
> On Tue, Jun 29, 2010 at 4:09 AM, John Baldwin <jhb at freebsd.org> wrote:
>   
>> On Monday 28 June 2010 1:57:27 pm Ireneusz Pluta wrote:
>>     
>>> John Baldwin pisze:
>>>       
>>>> On Monday 28 June 2010 12:00:06 pm Ireneusz Pluta wrote:
>>>>
>>>>         
>>>>> John Baldwin pisze:
>>>>>
>>>>>           
>>>>>> On Friday 25 June 2010 4:59:57 pm Ireneusz Pluta wrote:
>>>>>>
>>>>>>
>>>>>>             
>>>>>>> John Baldwin pisze:
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>> Hmmm.  You might have a hardware issue.  OTOH, you can try seeing if
>>>>>>>>                 
>> you
>>     
>>>> have
>>>>
>>>>         
>>>>>>>> a BIOS option to disable PCIE error logging.
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>> is it one of them?:
>>>>>>>
>>>>>>> Assert NMI on SERR
>>>>>>> Assert NMI on PERR
>>>>>>>
>>>>>>> (pdf page 109 of: ->
>>>>>>>
>>>>>>>
>>>>>>>               
>> http://download.intel.com/support/motherboards/server/s5520hc/sb/e39529013_s5520hc_s5500hcv_s5520hct_tps_r1_9.pdf)
>>     
>>>>>>>               
>>>>>> Well, that will turn off the NMIs.  Not sure if it will affect the event
>>>>>> logging, but it is worth a shot.
>>>>>>
>>>>>>
>>>>>>             
>>>>> Per BIOS setup documentation:
>>>>>
>>>>> On SERR, generate an NMI and log an error.
>>>>> Note: [Enabled] must be selected for the Assert NMI
>>>>> on PERR setup option to be visible.
>>>>>
>>>>> and:
>>>>>
>>>>> On PERR, generate an NMI and log an error.
>>>>> Note: This option is only active if the Assert NMI on
>>>>> SERR option is [Enabled] selected.
>>>>>
>>>>> However, disabling them did not change anything.
>>>>>
>>>>>           
>>>> Is it still logging errors and sending NMIs with them disabled?
>>>>
>>>>         
>>> with the options I mentioned disabled. They do not have to be the only
>>> sources of NMIs, do they?
>>>       
>> Well, they should be the sources of the log messages you found in your system
>> event log.  There is a good chance that you have some broken hardware
>> somewhere, I'm not sure how easy it is for you to debug that via swapping out
>> components, but the RAID controller is the first thing I would try.
>>     
>
> You might want to try a BIOS update to see if it resolves these
> problems ... I seem to remember some mention of these sorts of errors
> in the change log of one of the recent Intel server board BIOSes.
>   
Yeah, I tried, and the problem dissappeared.
I really could do that initially but none of the entries found in the 
BIOS changelog from Intel seemed to address my particular issue.