Can SSA or HACMP handle a disk replacement automatically

[Jeffrey Ross]

I'm running AIX 4.3.3 on a 44P connected to 2 SSA enclosures, each with 4
disks. The disks in one enclosure are mirrored to the corresponding ones in
the other enclosure using LVM.
Is there a way of setting up SSA so that a hard drive failure can be
resolved simply by pulling the faulty disk and inserting a new one? (Ie the
system would have to detect the failure, notice the new disk being inserted,
include that disk in the volume group, and resynchronise the mirror.)
I know how to do it manually but am wondering if it can be automated so that
a hardware technician can resolve the whole thing without my involvement.
If it can't be done with SSA does HCMP perform this function?
Thanks,
Jeffrey.

[Steve Nottingham]

"Jeffrey Ross" <jeffrey.rossATairways.co.nz@no.spam> wrote in message news:<3f727b81$1@news.iconz.co.nz>...

> I'm running AIX 4.3.3 on a 44P connected to 2 SSA enclosures, each with 4
> disks. The disks in one enclosure are mirrored to the corresponding ones in
> the other enclosure using LVM.
> Is there a way of setting up SSA so that a hard drive failure can be
> resolved simply by pulling the faulty disk and inserting a new one? (Ie the
> system would have to detect the failure, notice the new disk being inserted,
> include that disk in the volume group, and resynchronise the mirror.)
> I know how to do it manually but am wondering if it can be automated so that
> a hardware technician can resolve the whole thing without my involvement.
> If it can't be done with SSA does HCMP perform this function?
> Thanks,
> Jeffrey.

SSA stands for Serial Storage Architecture which is a disk subsystem.

In short, no this cannot be done automatically as AIX cannot recognise
the disk having been changed without running cfgmgr. You could of
course write a script to carry out all the work, but it would require
human intervention, to tell it when/which disk has been replaced.

Of course if you have a RAID-5 array with a hot spare, then it is
possible for the system to recognise a new disk.

Steve

[Andreas Schulze]

"Jeffrey Ross" <jeffrey.rossATairways.co.nz@no.spam> schrieb im Newsbeitrag
news:3f727b81$1@news.iconz.co.nz...

> I'm running AIX 4.3.3 on a 44P connected to 2 SSA enclosures, each with 4
> disks. The disks in one enclosure are mirrored to the corresponding ones

in

> the other enclosure using LVM.
> Is there a way of setting up SSA so that a hard drive failure can be
> resolved simply by pulling the faulty disk and inserting a new one? (Ie

the

> system would have to detect the failure, notice the new disk being

inserted,

> include that disk in the volume group, and resynchronise the mirror.)
> I know how to do it manually but am wondering if it can be automated so

that

> a hardware technician can resolve the whole thing without my involvement.
> If it can't be done with SSA does HCMP perform this function?
> Thanks,
> Jeffrey.
>

Hallo Jeffrey,
depending on whether you can define exactly under what circumstances a disk
should be made to leave the mirror it would be quite easy to write a script
that will do that for you. Adding and removing ssa disks to/from an LVM
mirror and replacing them physically can be done online. The script itself
should not be a problem as you said you know the steps manually. (If your
are unsure, search this group - all steps have been discussed in great
detail before). Such a script could then be used by HACMP of course.
However, there is no build-in function in HACMP that would do that for you
already. So the one and only problem is the handling of the error conditions
as some errors never hurt, some errors hurt when they occur too often only
(e.g. DISK_ERR4) and some errors lead to a complete hardware breakdown (that
must be distinguished from a temporary power loss of a single disk / a
complete drawer / a node ...).
Happy hacking!
Andreas