Server RAID - HDD failed

[mcloum]

Evening all,

So it was bound to happen, i have a 4 year old server with a Ultra 320-1 RAID card in. 1 of the drives failed tonight, its been years since i've done anything with RAID.

Question, does the replacment drive have to be the same make/model? same speed and roughly the same size? I assume if its larger it will just use what the other drives are (70gb)

Sorry to sound daft, like i said its been years.....

Cheers

Mike

[sammy]

Hi Mike

yes the replacement drive ideally should be the same size and speed, cant be smaller in size, but it can be larger....but if it is larger or faster then the additional capacity and speed will be wasted becuase the RAID set only utilises the space equal to the smallest disk in the set, and the speed equivelent to the slowest speed in the set of disks.

I can imagine a scenario where you are using old 36gb disks and now you are in a situation becuase its hard and very expensive to get 36gb disks and therefore you need to go for a new 72gb disk. In this situation you just have to bite the bullet, and use a 72gb disk.

Alternatively you could use this as a opp to backup all the data, destroy the raid set, and recreate with bigger disks alll round.

choice is yours.....hope i didnt ramble too much above :-)

good luck.


Sam

[teejay]

As long as its larger it will be fine. The problem arises when you get a different model/make of supposedly the same size as they can actually be slightly different and if they end up being fractionally smaller then you can't use it. Also, if it is a slower hdd then it will degrade the performance of the raid array.
Another tip, if you have space to fit it, don't take out the old drive straight away. Add the new drive as a hot spare and leave for about a day and it will automatically rebuild onto the new drive. Once this is done, take out the dead drive. I've found this method is usually a lot easier and more successful.
Also, as the server is quite old now, if you intend keeping the server for a while order an extra drive and put in as a hot spare once its rebuilt, with all these things when one drive goes its sods law that another one will shortly follow!

[yabbadabba]

Agree with Teejay, once one drive fails the others are often not far behind.
The last company I worked at, once 1 x drive failed we would swap out all of them with new drives.
The reason for this was we had a spate of RAID failures over a year, an engineer would make a site visit, replace the faulty drive, come home, go to the next site and often within 2/3 weeks another drive has failed.
As we had sites all over the UK is was more cost effective to replace all rather than drive back too the same site, especially when in Perth or Glasgow and our company was based down South

[mcloum]

Hi all,

Thanks for all the replies, it gives me something to think about. BSF did talk about it being a problem if the hardware fails at this point, i replied "been ok so far " Sods law it had to happen!

I did take the drive out and put it back in and it started rebuilding but im not hopeful it will sort it out. the BIOS (at boot) did report the drive a degraded, not failed?

I think i will approach this as worst case scenario and restrcuture the server. Put in a IDE drive for a system drive and use new scsi disks for the RAID. At the moment its got 3 73gb drives with 2 partitions, 1 syste,m and t'other data, not ideal

Cheers guys...any other ideas and opinions greatly appreciated.

Mike

[teejay]

If you are that close to BSF it seems a bit pointless reconfiguring the server. I would just buy in a spare drive and put it in as a hot spare, then if the dodgy drive does fail you'll be ok.

[mcloum]

Yes i guess so, but we are not an academy for another year, and then not in our new building until about 2011 (which wont happen due to the land they want to build on being flooded.....twice since the 2007 floods)

Update: Came in this morning and the server is no longer failed, green across the board. Wonder how long it will hold out?

[Midget]

until someone saves a single copy of something VERY important in that magical time between backups and the drive failing.

[psydii]

12 months off our BSF transition and we bought servers to replace ones that were old/out of warrenty/at capacity.

Can you afford not to?

If so, just replace the failing disk, and then the good one - to be sure!