Unsolved
This post is more than 5 years old
25 Posts
0
1654
February 21st, 2012 04:00
Add luns to Storage groups --> I/O Interruption
Hi all,
I am just trying to explain some complaints I received today from the database administrators...
Some luns were not able to service I/O at a particular time... I checked the CX, vSphere and the Windows virtual machines, and they were not reporting any error. Performance was also not degraded.
At the same Oracle reported an error, I was modifying the Storage group for that Oracle cluster and added some LUNs.
Can such action result in I/O errors on database systems? In my opinion luns can be added with no interruption of I/O service. Nut maybe that isn't true for Storage Groups. presenting multiple (60+) luns to hosts.
Regards, Bart
No Events found!
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
February 21st, 2012 06:00
i've never experienced service interruption during storage provisioning. If all you were doing was adding new LUNs and did not even get to the part where the host was rescanning to see them ...i don't see how it could cause issues. Were there any errors in PowerPath , syslog ?
SKT2
2 Intern
•
1.3K Posts
0
June 3rd, 2012 08:00
are you saying adding LUN is the only change you did to storage groups? what OS is this for oracle cluster? any device scanning tool ran on the host side immediatly after presenting LUN?
bartdonders
25 Posts
0
June 4th, 2012 01:00
Hi SKT,
To be more specific, the oracle cluster is an vSphere cluster, hosting virtual Windows Server 2003 or Windows Server 2008 R2 virtual servers. There is no clustering on Microsoft level.
I added a few luns to the ESX cluster. On that moment, disk errors were reported by multiple virtual servers.
I am not convinced the problem is in the CX, but I just wanted to check that in this thread.
Regards, Bart
ashes1
8 Posts
0
June 5th, 2012 13:00
I have seen something of that sort long ago but it was specific to an earlier release of flare 30 with virtual provisioning only.
You would see in the SP event logs events from virtual provisioning MLU driver, they may appear as normal but the drive experinced problems.
I do not remember the specific flare version, but the MLU driver had memory leaks and also had a SOC memory limit ( safe operating capacity). It has since been fixed though.
If you search for SOC on powerlink, you may find sveral primus cases.
rupal1
433 Posts
0
June 6th, 2012 07:00
May i know the flare on the array. I think you might have to open a case with the support to get the logs analysed accordingly.
Rupal Rajwar