2 Disks always blinking in the enclosure

Taratikos's picture

We have 2 Mac Pro 10.8.2 , Xsan 3.0, with Atto fiber 8g acting as MDC and MDC Failover.
We have 2 LUNS,(metadataandJournal and DATA).
We recently faced a sudden and strange issue.
The performance suddenly degraded to half.
We noticed that 2 specific drives are always blinking even of no one is accessing the SAN.
the disks are healthy and the volume is online and there is no issue in ATTO FastStream.
When we stop the volume in Xsan, the blinking stops.
We are intending to upgrade the XSAN admin to 2.2, but till then, Has anyone faced such issue??

JSamuel's picture

Which RAID set do the 2 drives belong to... Are they RAID1 set for your metadata LUN?

Joel Samuel.
/thirtytwo - Apple Consultancy & Direction
Proud sponsor of Xsanity.com

All contributions are my own personal opinions - not those of any entity I represent.

Taratikos's picture

First we upgraded to the latest OSX Server and the problem persisted.
The 2 disks ( 1 and 13 ) are blinking since the creation of the Xsan Volume.
The metadata Lun and the data Lun were created through the raid controller.
So the metadata Lun is part of all the 16 disks in raid 5.

It has been 6 months that we are working normally.

JSamuel's picture

Sure you havn't got the ATTO utility doign an "identify drive" for you?

Also, your metadata LUN should be a different RAID set, typically a 2-disk RAID1, your data LUNs can then be RAID5 or RAID6 sets etc.

Joel Samuel.
/thirtytwo - Apple Consultancy & Direction
Proud sponsor of Xsanity.com

All contributions are my own personal opinions - not those of any entity I represent.

Taratikos's picture

The drive blinking is related to Xsan. Nothing related to Atto.
When we stop the Volume in Xsan Admin, the blinking stops.
You are right about Raid1 for Metadata, but we did our own configuration for the LUNs and it worked.
But Now it seems that the MDC is always using the Metadata Lun to write data, since in Atto, we can see that the disks activity is always in Write.
We did all the maintenance commands for the SAN ( CVFSCK, SNFSDEFRAG...).
And they showed No Errors.

xsanguy's picture

How did you measure the performance drop? How did you first notice it? Did anything else change around the same time? What is your application?

How much RAM do your MDCs have? How full is your SAN? I've seen this on very full very active SANs with insufficient MDC memory.

I have experience with the Faststreams, how do you have them setup? All LUNS exposed to all ports? How is your FC zoning? OD/AD Bound?