[OmniOS-discuss] Parity error on path mpt_sas2

[Dan McDonald]

> On Nov 25, 2014, at 10:00 AM, Filip Marvan <filip.marvan at aira.cz> wrote:
> 
> Hi,
>  
> I have OmniOS 12 with three RAIDZ pools.
> In /var/adm/messages a have these messages about every few seconds:
>  
> Nov 25 15:30:52 servername scsi: [ID 107833 kern.notice] /pci at 0,0/pci8086,e08 at 3/pci1000,3020 at 0 (mpt_sas0):
> Nov 25 15:30:52 servername      Aborted_command!
> Nov 25 15:30:52 servername scsi: [ID 243001 kern.warning] WARNING: /scsi_vhci (scsi_vhci0):
> Nov 25 15:30:52 servername      /scsi_vhci/disk at g50014ee209cf0116 (sd13): Parity Error on path mpt_sas2/disk at w50014ee209cf0116,0
> Nov 25 15:30:52 servername scsi: [ID 107833 kern.notice] /pci at 0,0/pci8086,e08 at 3/pci1000,3020 at 0 (mpt_sas0):
> Nov 25 15:30:52 servername      Aborted_command!
> Nov 25 15:30:53 servername scsi: [ID 107833 kern.notice] /pci at 0,0/pci8086,e08 at 3/pci1000,3020 at 0 (mpt_sas0):
> Nov 25 15:30:53 servername      Aborted_command!
>  
> In this log messages there is error on sd13, but there are similar parity errors on four other disks too. All disks with parity errors are in the same RAIDZ pool. 
> All disks seems to be OK in zpool status and there are no errors on disks reported by smartctl. There are no problems reported by fmdump 
>  
> All three pools using the same LSI 9211-8i HBA with firmware version 20.
>  
> Sometimes that pool have very slow responses even on "zpool status" command, sometimes works normaly.
>  
> Do you have any idea where could be a problem? What else I can check?
>  

You're not being clear enough.  Is sd13/g50014ee209cf0116 the only disk reporting errors?  If so, replace it with a new disk.  If there are other disks reporting errors, you might want to replace them as well.  It looks like the SCSI commands are failing, but perhaps succeeding upon retry.  I'd check your cabling and see which disks are reporting the problems.  If you bought them in batches, maybe you have a bad lot?

Dan