Performance/stability problems on PowerEdge 1950

Douglas K. Fischer fischerdk at fidoki.com
Sun Dec 17 19:21:08 CST 2006


We've run into the same issues going from 2850s/6650s to 2950s. Running
identical I/O loads perform more than an order of magnitude slower on
the PERC5/i with SAS drives than PERC4 with SCSI drives. We've also
migrated 1850s to 1950s, but their roles are not I/O intensive so we're
not seeing any issues there (yet).

We've verified PERC card configurations, tried updated BIOS, PERC
firmware, and megaraid_sas drivers (from Red Hat, Dell, and LSI), all to
no avail.

One of my guys is banging his head against the wall on this. There are
so many possible variables (driver parameters, kernel parameters, driver
bugs, PERC firmware issues, etc). Hopefully if enough attention comes to
the issue there might be some more ideas floated around to try...

Doug

Dean Manners wrote:
> Harry,
> 	I have an identical anecdote, coming from 1850 to 1950.  Under high
> disk I/O the machine becomes very laggish and generally performs badly in
> comparison to the SCSI based 1850's.  Running Debian with 2.6.18, moduled
> megaraid_sas, sd_mod, scsi_mod. Also keen to hear of any ideas or
> directions..
>
> Regards
> __________________________________________ 
> Dean Manners
>
>
>   
>> -----Original Message-----
>> From: linux-poweredge-bounces at dell.com 
>> [mailto:linux-poweredge-bounces at dell.com] On Behalf Of Harry Roberts
>> Sent: Thursday, December 14, 2006 11:05 PM
>> To: linux-poweredge at dell.com
>> Subject: Performance/stability problems on PowerEdge 1950
>>
>> Hi all,
>>  We recently upgraded from SCSI based Dell 1850s to SAS based 
>> Dell 1950s and found that under the same loads, stresses, 
>> applications & conditions the SAS based 1950 system starts 
>> failing and becoming unresponsive at a much earlier point.
>>
>>  One of the main roles of the servers are to perform 
>> telephone call recording, which in turn means that there are 
>> lots of concurrent writes per second (say 20 per second, per 
>> concurrent call). The SCSI based system was able to easily 
>> handle 60+ calls, while the SAS based system was having 
>> serious problems at around 30.
>>
>>  We also performed the same tests on a SAS HP DL360 g5 
>> (SmartArray RAID controller), which performed even worse and 
>> stopped operating normally at around 7 concurrent calls.
>>
>>  On both we're using Gentoo with the stock 2.6.18.x kernel 
>> and Gentoo's patched 2.6.18.x kernel, we then tested again on 
>> a 1950 server using Redhat EL 4, and none of these problems 
>> occurred. It seems that something in Redhat's heavily 
>> modified 2.6.9 kernel is fixing it - but we can't track down 
>> exactly what.
>>
>>  Has anybody run into similar problems with system load 
>> skyrocketing or application response times increasing 
>> massively when there's high disk load on SAS any of Dell's 
>> SAS based PowerEdge series servers?
>>
>>  Any ideas or possible directions we could go in are appreciated.
>>
>>  Thanks,
>>
>> - Harry Roberts
>> - Software Developer, Telappliant Ltd.
>> - Phone: +44 (0)207 043 3519
>> - Email: harry at telappliant.com
>>
>> --
>> No virus found in this outgoing message.
>> Checked by AVG Free Edition.
>> Version: 7.1.409 / Virus Database: 268.15.18/586 - Release 
>> Date: 13/12/2006
>>  
>>
>> _______________________________________________
>> Linux-PowerEdge mailing list
>> Linux-PowerEdge at dell.com
>> http://lists.us.dell.com/mailman/listinfo/linux-poweredge
>> Please read the FAQ at http://lists.us.dell.com/faq
>>
>>     
>
> _______________________________________________
> Linux-PowerEdge mailing list
> Linux-PowerEdge at dell.com
> http://lists.us.dell.com/mailman/listinfo/linux-poweredge
> Please read the FAQ at http://lists.us.dell.com/faq
>   



More information about the Linux-PowerEdge mailing list