0

I have a server with IBM ServeRAID 8k-l RAID controller. It has two HDDs in RAID 1 array. After the server reload, IBM ServeRAID 8k-l RAID controller started the rebuild operation for physical drive #1:

# arcconf GETSTATUS 1  
Controllers found: 1                           
Logical device Task:                           
   Logical device                 : 0          
   Task ID                        : 101        
   Current operation              : Rebuild    
   Status                         : In Progress
   Priority                       : High       
   Percentage complete            : 3          


Command completed successfully.                
#

During that time the virtual machines in this physical server logged file-system related errors. I disabled the physical drive #1 and thus stopped the rebuilding process and after that, the virtual machines booted up.

Was this an expected behavior? Should the IBM ServeRAID 8k-l RAID controller performance be degraded during the rebuild operation?

  • 3
    Performance (speed) is degraded, but you shouldn't see any errors. – Halfgaar Oct 5 '17 at 13:31
  • @Halfgaar Ok. However, am I correct that at the time of "Rebuild" operation the ServeRAID does not use this physical disk for reading and write operations? – Martin Oct 5 '17 at 14:32
  • 1
    Under normal circumstances, it's smart enough to know which sectors can be read/written from/to both disks and which aren't. In your case though, something may be wrong that we're not seeing, since a drive was kicked out, and you're seeing errors on the application level. – Halfgaar Oct 5 '17 at 18:10
  • @Halfgaar I removed this problematic physical drive from server and tested it with SMART self-tests and badblocks. Turns out, that drive has I/O errors. I guess that after the server reboot RAID controller considered this problematic physical drive as a new HDD and thus started the rebuild operation? In addition, looks like despite the rebuilding process, the controller already used the problematic physical drive for reading and writing data which caused errors on the application level? – Martin Oct 15 '17 at 15:32
  • I suspect that the other drive may also have errors. Make sure that the rebuild succeeds. You may then want to check SMART for errors on the other disk. On some controllers, you can do that while it's part of the RAID array; no need to remove the disk. Check the smartctl man page. – Halfgaar Oct 15 '17 at 17:13
0

You can change the priority of rebuild operation (task) to LOW or MEDIUM. LOW priority will allow to increase logical disk performance during the rebuild operation. To setup default value for tasks use command:

arcconf setpriority 1 LOW CURRENT

To change the priority for running task use command:

arcconf setpriority 1 <Task ID> LOW
  • However, even with "High" rebuild operation task priority, one should not see errors on OS level? – Martin Oct 15 '17 at 15:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.