megaraid_sas waiting for command and then offline

Joseph Malicki jmalicki at metacarta.com
Mon Dec 11 23:58:39 CST 2006


David Gwynne wrote:
>
> I'm experiencing this (or something that sounds extremely similar to 
> this) on a PowerEdge 2850 with a PERC5/E running Solaris and a driver 
> for the perc I wrote. IO runs beautifully until a couple of commands 
> are submitted that are never processed by the controller. After that, 
> IO is blocked and then I have to powercycle the machine (reboot waits 
> for IO to finish).
>
> I'm able to reliably reproduce the problem, which is very annoying 
> because I want to use the machine running solaris in production.
>
> I have the PERC5/E hooked up to an MD1000, which is populated with 15 
> 500GB SATA disks. The disks are configured into a RAID50 (3 raid fives 
> of 5 disks each).
>
> The code for my driver is up at https://svn.itee.uq.edu.au/repo/mfi/ 
> if anyone wants to play with it.
>
> Is there a way to disable patrol read from the controllers bios, so I 
> can try to see if that affects the reliability of the controller. 
> Obviously I can't modify that setting from within the operating 
> system... I'd love to get hold of some doco ;)
>
> dlg
>
Hi Dave,

Thanks for the quick response, it's good to know we're not alone.  Do 
you have a workload that you know reproduces it (we haven't been able to 
find one thus far, though the different configuration may be related).  
What hardware are you running, if we can narrow this down (beyond 
mentioned above):

 1. Are you running SMP/what chips? (we have dual Woodcrest 5160, which 
are both dual core chips)
 2. How much RAM do you have (which may implicate or exonerate any 
issues with DMA >8GB)?  (we're running 16GB)
 3.Which PERC 5/E firmware/package version are you running? (we have 
been using 5.0.1-0003 when we've seen the problems)  does 5.0.2-0003 help?

I've only ever accessed patrol read etc. settings through the MegaCli 
utility for Linux, so I'm not sure if you can through the BIOS (though I 
would assume so?).

Are you running the firmware below, or the 5.0.1-0003 version?

Thanks!
Joe
>>
>> Brett, is this still happening to you?  We're seeing this very
>> sporadically, but it does concern us.  We've seen driver updates in
>> 2.6.19 (v00.00.03.05) and a new Dell PERC 5/i firmware:
>>
>> Package Version - 5.0.2-0003
>> Firmware Version - 1.00.01-0157
>> SASBIOS Version - MT23
>> Ctrl-R Version - 1.02-007
>> MPT Version - 00.06.71.00-IT
>>
>>



More information about the Linux-PowerEdge mailing list