Unsolved
This post is more than 5 years old
60 Posts
0
2180
cx4-480 clone causing multipath errors on RHEL 6
On our cx4-480, we have a clone that is mapped to a LINUX RHEL 6 server. The source of the clone is a Windows server. When the clone is syncing, we are getting multipathing errors and utilization on the box is high, when the clone is fractured, we do not get the errors and the utilization goes back to normal.
When this LINUX server was at RHEL 4, we did not experience this issue, only since we've gone to RHEL 6. The CLARiiON connectivity setup has not changed (ie. failover mode was the same), the only thing changed is the level of Red Hat.
Has anyone experienced anything like this? EMC Support is shooting blanks at this point.
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
0
October 22nd, 2013 06:00
getting error messages regardless if the clone target is mounted on RHEL6 box or not ? ( i am curious what you do with NTFS partitions on a RHEL box, raw backups ?)
gem_guy
60 Posts
0
October 22nd, 2013 07:00
The clone is not even mounted, the error messages are coming when the LINUX is doing a health check on the lun.
We mount the NTFS on the LINUX box and back it up.
gem_guy
60 Posts
0
October 22nd, 2013 07:00
The high utilization is on the LINUX host, not the SP's.
anand2405
20 Posts
0
October 22nd, 2013 07:00
multipathing errors would be seen, cause when you put the clone in syncing it would not be accessible by the host due to which you might see multipathing errors.
do you see more of cache utilization or does the SP utilization shoots up? did you get the NAR files analyzed with EMC support?
Regards
Anand.
kelleg
4.5K Posts
0
October 25th, 2013 09:00
When a CLone is in Sync mode, it is NOT accessible to a host, only when you fracture a Clone can you access it.
Please see the attached see page 7 "Data Access" section. In the document, BCV refer to Clones.
glen
1 Attachment
Using SnapView for Business Applications.pdf
gem_guy
60 Posts
0
October 25th, 2013 10:00
glen,
Thank you for your response and pdf.
I understand that the syncing clone would be unaccessible by the host, but would you expect the LINUX host to be getting multipathing alerts from the clone lun while it is syncing?
Or to ask it another way, would I be expected then to remove the clone from the host storage group on the array while it is syncing, and then add it back to the storage group, do a rescan on the host, then be able to mount onced it's fractured.
gem_guy
60 Posts
0
October 29th, 2013 10:00
glen -
I don't know if we are mixing our terms or not. We are getting path errors while the clone is not mounted (but mapped to the storage group on the CLARiiON).
On our HPUX systems, the clone can stay mapped to the host storage group while it is synchronized, then we fracture the clone, mount the clone, backup the clone, unmount the clone, then wait until the next synchronization. We do not have to unmap the clone from the storage group on the CLARiiON.
Apparently from EMC, this is not the standard practice. According to them, the clone should be unmapped from the storage group, then synchronized, then fractured, then added back into the storage group, scan the IO bus to pick up the clone, mount the clone, backup the clone, unmount the clone, remove the device file, unmap the clone from the storage group, wait until next synch.
kelleg
4.5K Posts
0
October 29th, 2013 10:00
Yes - if a disk becomes unavailable to a host, that host will complain about it. If the Clone is syncing it is unavailable. Unmount the Clone, then sync, check state/status to ensure the clone is in Synchronized state, fracture, then mount.
glen
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
0
October 29th, 2013 11:00
agree, unmount should be sufficient. We do this every day the only difference my hosts that are part of this process are on VMAX not Clariion/VNX.
gem_guy
60 Posts
0
October 29th, 2013 13:00
Are there any LINUX users, with CLARiiON disk, specifically mounting and backing up clones?
Are you experiencing the same issue?