Removing 1 disk and replace it from a Raid LUN.

francisyo · Xsan Master Joined: 06 Nov 2006 Posts: 80

Hi all,

Its been a while since my last post. I am now facing a big problem in which i haven't experience the solution. Here is the situation:

We have 2 metadatas and 2 Xsan volumes. We are already in Snow Leopard 10.6.7 and our Xsan is 2.2.1

Our Xsan volume is almost FULL and one of my Raid Lun is already alarming even on and after I Reconditioned it.

Disk 2 Reported an Error. Command 0x25 error: 0x40 STATUS..........

I am planning to replaced this disk with a new one but I am really afraid there will be problems with my Xsan Volume which has 52TB of files in which it has no backup.

If anyone out there who successfully done performing this process of Replacing Disk on a Raid without any error on the Raid and Xsan please do tell me.

Can you tell me the step by step procedure. I really really need it.

Thank you very much in advance.

lucasnap · Xsan Master Joined: 05 Oct 2006 Posts: 107

Hi francisyo,

I'm guessing you are using Apple's Xserve RAID's because of the type of error. This error looks like bad blocks. I'm also guessing that you don't have hotspares in the RAID system. So here is what you want to do.

Step 1: Make a backup
Step 2: Replace Disk 2 with a new disk, same size or bigger. Of course this should be a PATA (IDE) disk, not very easy to get these days. I hope you bought some spares a while ago.
Step 3: The mirror should start recovering automaticlly, if not you can check this in the RAID admin.

Keep in mind that the Xserve RAID are EOL a long time ago, IDE disks are not made anymore, so I should think about replacing your hardware.

good luck, lucas

Solidus · Xsan Master Joined: 28 Jan 2011 Posts: 70

Just to add to that good advise.

Try to do the rebuild during the evening on a Friday if you can or a suitable time when the SAN is not under load as it will speed up the rebuild.

francisyo · Xsan Master Joined: 06 Nov 2006 Posts: 80

Hello Lucasnap & Solidus,

Thank you very much for the reply. I have a hot spare on all my LUNS. Instead of striping 7 disk, I made it 6 disk stripe and 1 hot spare. I been doing this since Xsan 1. But my only problem is I haven't tried replacing 1 disk on 1 LUN. I have also some spare 750gb of PATA disk for replacement. I am just not eager to replace the problematic disk on my raid in fear of loosing everything. My boss is still on the process of trying to solve the backup problem. So meaning it is still a LIVE one with an irritating alarm sound every now and then. By the way its a Raid 5 with 1 hot spare.

abstractrude · Xsan Master Joined: 13 Mar 2008 Posts: 860

There are other things you can do here. why not just dump journal before replacement?

stop your volume. dump . replace disk. finish rebuild. start volume. if it doesn't rebuild properly.

snmetadump is a far better plan than what you are doing now. hope this helps.

francisyo · Xsan Master Joined: 06 Nov 2006 Posts: 80

Hello abstractrude,

I am not very good with terminal commands but thank you sir for this advice. And thank you to all who replied. Ill gather all your tips and soon when im confident enough i will perform this.