We are on HACMP 4.3, the disks numbering is out because of different numbers
of internal SCSI disks. This was made worse because I recently had our IBM
tech add some more SCSI and SSA disks to the two machines. I was aware of
the dummy disks id's but was kind of wondering why they looked to be a waste
of time - I didn't realise that restoring from a mksysb would re-number
everything. I have progressed since posting y original message in that I
have gone through SSA device screen on SMIT and identified all the physical
disks associated with logical disks on both machines. I figured a light
flashing was going to eliminate any ambiguity.
I have two nodes, Node B is production and working fine. Node A is standby
and working fine except it doesn't have the VG's from NodeB defined on it
anymore which will cause me grief in the event of a failover. Using the
importvg command on node A passing as a parameter a physical disk that is a
member of a VG on Node B, the system is saying the disk is not a member of
any VG. I did a varyonvg -u -b on Node B before doing the importvg on A, I
also passed the major number and the VG name and the diskname on A to
importvg.
I don't think there is a major problem here, I think it is just a crucial
step that I am missing out. Any ideas on what it might be? Do I need a
synclvodm command run on one or both my nodes?
Thanks
Bernie.
>If you're at HACMP 4.4, you should be able do this via "smitty hacmp ->
>Cluster Systems Management -> Cluster Logical Volume Manager Shared
>Volume Groups -> Import a shared volume group".
>If not, well, you want the names to be the same as on the other machine,
>so run lspv on the other machine and then look up the PVID in the lspv
>output on this machine, get the corresponding hdisk number . And
>remember, you only have to find one hdisk per VG, not all of them.
>OR, you could figure out why the numbering is different and how -
>usually it's because the two servers have differeing numbers of internal
>disks. The traditional solution for this is to add a "fake" hdisk, which
>is not really there as a SCSI id that's not there to even things up.
>Or did you restore a mksysb onto both machines? In which case, the
>choice is yours.
>> Hi all,
>> I'm new to AIX and am having trouble with a two node HA cluster.
>> I recently had to restore from a mksysb tape. Prior to the restoration,
the
>> hdisk numbers on the two nodes were syncd, that is hdiskx on node A
>> corresponded to hdiskx on node B.
>> After the restoration, hdiskx on node A no longer corresponds to hdiskx
on
>> node B, in fact there seems to be no relationship other than the serial
>> number on the disk.
>> Is there a recommended way of finding out which disk on node A maps to
>> which disk on node B? I have 12 volume groups to re-import and around 50
>> disks on each machine. Finding which is which is a little daunting.
>> Thanks,
>> Boris.
>--
>comp.unix.aix FAQ:
>http://www.faqs.org/faqs/by-newsgroup/comp/comp.unix.aix.html
by