Disks in Raid 5 on ML370

[spi200]

Hi all

I need some help with this one.
We decided to expand our Raid 5 on an ML370 and replace our 9.1Gb Ultra2 10K SCSI drives with 36.4Gb U320 15K SCSI Drives. I have replaced One drive at a time allowing to rebuild etc, but after the 2nd drive was added the Server abended overnight. Server had been very reliable and the only change has been to our disk config. Maybe the drives are having trouble with 4*10K & 2*15K drives or the Ultra2\Ultra3 combination in the array. Once the larger drives were in I was going to run CPQONLIN to expand the DATA volume. Any thoughts ? Is this allowed and will it work or am I on thin ice.
Any help would be much appreciated while I await my next abend.

Regards

David Dundas
System Administrator

 

[terry712]

should be a problem
i've used this method a few time
what model of 370 it it
i assume it's g2 or 3
what's in the abend log - or better in the cpqiml
i assume the latest nssd is on it
should be 7.01
you can tell by typing
m cpq*
give me a few of the result and i will know

cpqonlin will not expand the data volume - it can only expand the array
then nssmu will expand the data
assuming it's nss and not traditiona;l

 

[terry712]

sorry shouldnt be a problem

 

[spi200]

Thanks Terry172

Yes, I have used CPQONLIN a few years ago to expand an array before, but I had not mixed Ultra2 and Ultra3 SCSI drives before. I have since confirmed with HP that this is OK. Also in Netware 5 you can add the new volume segments to an existing volume. ie Data volume.

Regards

David

 

[technome]

Two cents...
If these are Seagate drives, there was a firmware problem with drives with ver. <6.0. with 10k and 15k drives, which might cause this. Nasty part, Seagates Enterprise manager destroyed a raid on me in the flash upgrade process.

 

[spi200]

OOUCH

I swapped out all 6 drives one at a time over about a week with no problems. I ran CPQONLIN to create a logical partion on the free space with no problem. I then decided to give the server a clean re-boot before adding memory or extending the Netware volumes etc. The Server never restarted, No SYS volume. It had a DATA volume which needed which repair would see and found lots of errors. Log a call with Novell support and they said I was stuffed and suggested to rebuild the Server. 22 hours later I had something of a server again, but will take a while to fine the thing. What an experience in DR, I won't go swapping drives again into old hardware.
The new Server build seems quiet fast now with extra RAM and 15K drives.

Regards

David Dundas
CNE

 

[technome]

Dave,
This bring up an incident I had years ago with a server where the transaction file grew to the full size of the remaining space, due to a zip file from Veritas. I was completely lock out until I obtained a util from Novel to delete the file. Kind of the same M.O

 

[terry712]

out of curiosity - was the build of this from 6.1 to 6.3

as there was an known issue with this build if the dos partition was a certain size that on reboot you wouldnt see the sys

there was a patch for it

certainly i known the disks shouldnt have been an issue as certainly i have used this method dozens of times

 

[spi200]

Hi Terry712

The Netware build was v 5.1 SP6 and I was only extending the data volumes because we were out of space. I to have done this method a number of times but never again.

Regards

David.

 

[terry712]

sorry i mean the compaq smartstart build
was it from 6.1 area

 

[spi200]

Hi Terry

The smartstart version would have preceeded that, probably early 5.x. I rebuilt the server after it crashe with 5.5 as this is the latest that support the ML370G1.
Novell support were most helpful and worth the $450US in the rebuild process.


Regards

David