TonyFlow

In that case, does it have an fsck command? If so run it on the volume that has the error, and it should tell you of any errors. Also, if you run a lsvg on the Volume group the disk is in, see if there are any stale PP's, if so, even after after a fsck, then it is a good indication that one of the disk in the raid is knackered

NeilT

It depends what level of RAID you are running, i.e. wether its mirrored, striped or spareset.

Usually if the disc is dead (and it doesnt sound like this one is) you can just whip it out and bung a new one in and hey presto 10 mins later its all ok (actually it never seems to work when I've done it!)

hope this helps

jase555

Hi Guys Wonder If You Can Help, New To The Alphaserver Scene And Have A Drive With An Orange Light Instaed Of Green, Me Thinks It Is Knackered, The System Is Set Up For Raid,

Does This Mean I Can Just Eject The Drive An Replace And The System will Sort It Out.

Thanks In Advance

Jase

ps Alpha server 1200 with 2 pedestals 8 (9.1 gb scsi drives)

JoeyDeacon

I have to admit my knowledge of raid is pretty shaky but I seem to remember something about Hot Swap Raid. This means you can replace one of the drives while the system is on and there is enough information on the other drives to rebuild the replaced one.

If it is not hot swap then DON'T just replace it without turning it off!!!

TonyFlow

On AIX systems (and a couple of NT ones I have used), the orange light signifies that the disk is being used)It is when just the orange light is on (no green one), that there is a problem, if the disks are RAID5?, then they are hot swapable, and all you should need to do is pull one out and put another one back in, and the disk will rebuild itself from data held on the other disks.
I am not sure how this translates to Alpha Servers, but if you run a defrag/disk checking utility, any errors on the disk should be reported!

ca

jase555,

It is also worth considering taking a full backup of the disk before you do anything, just to be on the safe side.

Is the box running OpenVMS by any chance?

C

jase555

Its running

digital unix and an oracle v734 database with an erp application

Ian Cook

Hot swap depends on the hardware not the raid set used, if the hardware supports hot swap then u can do it, if it doesnt u stand a chance of losing the whole array ! doesnt matter whether its Raid 0 or Raid 10 !

Raid 5 can rebuild itself if you have assigned a hot swap disk, but this is usually already in the array, raid then rebuilds the array onto the spare, making the faulty drive redundant ! then if its hot swap capable you can swap it out, if not you will have to power off the server to change the faulty drive. If you dont have a hot swap, then make sure the system can do hot swap before just yanking anything out !!!

Paul Frank

Message from the Gods of ERP application support here . . .

DON'T just pull it out (oo-err missus)
1) Stop the ERP application
2) Stop Oracle
3) do FULL backup (that's F U L L )
4) we KNOW it says hot swapable on the lid of the box BUT . . . .
5) power off
6) now swap disk

alternative strategy:
1) swap disk
2) cancel all social activities for next 3 days & nights
3) log on to jobserve

your call . . .
Paul

TonyFlow

Is there a RAID manager software (DAS), if so then the rebuilding side will be done through this! When I used to work in an AIX Oracle/SAP environment, b4 anything like this was attempted we shut down AP/Oracle, made 2 full backups, and 2 mksysbs (just in case), as all the drives were hot swapable, we kicked everyone off the system, unmount/export the volume group, plugged in the new drive, and then let the system rebuild (which for an 18Gb drive would take approx 4 hours!).
When this was finished, the data was checked for contingency, and then the system was rebooted, and if all was OK, then the system would be backed up again, if not, then the only thing to do would be a restore!!!

Afte