replacing failed drive DL380G7

[BradV]

I have two dl380 G7s with 8 2.5" 600GB dual SAS disks.  All the disks are set up in a single disk array.  On both systems, the disk in slot 3 failed.  I just had HP service out to replace the disks.  I'd like to be able to get the drives back online without having to reboot the systems (RHEL 6.2).  Does anyone know if I can use hpacucli to recreate the single drive arrays?  I was looking at hpacucli ctrl slot=0 array C delete, but it tells me it will wind up re-lettering the existing arrays.  There is active data on each drive.  I didn't do it because I wasn't sure how it was going to effect the file system.

Any suggestions would be greatly appreicated.

[Typhoon87]

The drives on that unit are Hot swapable. The big deal is what kind of raid you have. Is it a raid 5?

Do you have one drive configurated as a hot spare?

When the HP tech replaced the drive did he shut the machine down and you have not done anything else to it since?

[BradV]

No, each drive was a single array and no, we did not shut down the system.  Both are part of an hdfs network and it is up and in production.  I wanted to try and avoid having to take out one of the nodes just to replace the drive.  I was going to try and use hpacucli to remove the single disk array and then re-create it.  However, when I issued hpacucli ctrl slot=0 array C remove it replied that it would re-letter all of the remaining drives and asked if this was OK?  I wasn't sure how that would effect the current running disks.  So, I replied no.  The service tech is coming back out this morning, but he's just a hardware guy.  Doesn't really know about hpacucli or such.

[BradV · Accepted solution]

OK, the HP techs informed me that I would have to reboot each server in order to get the replaced drives into an array without effecting all the rest of the arrays.  So, I guess I can't do it live.  :(

Thanks!

[Torsten.]

So each hard disk is an individual RAID0?

If the disk has failed, this logical drive has failed too, of course.

From the hp-ux utilities for smartarray controllers i know this command:

# sautil device_file accept_media_xchg logical_drive_number

this makes such a failed RAID0 logical drive available again after the HDD is replaced without a reboot (of course without the data in it), so I assume you can trigger the same behaviour with other tools as well.

isn't it something like this example:



hpacucli> ctrl slot=0 ld 4 modify reenable forced

(make sure to use the correct controller and logical drive #)

[BradV]

That's great!  I was able to make it work without having to reboot the servers.  I did the following:

hpacucli ctrl slot=0 ld 3 modify reenable
parted --script /dev/sdc mklabel msdos
parted --script /dev/sdc mkpart primary ext3 0% 100%
parted --script /dev/sdc print
mke2fs -m0 -t ext4 -v /dev/sdc1
e2label /dev/sdc1 had3

 Thanks for the suggestion!