From: Green, Simon (SGreen@KRAFTEUROPE.COM)
Date: Thu Jun 20 2002 - 06:14:39 EDT
I've worked with shared loops, both in HACMP and non-HACMP environments.
For the HA environment - and this will apply to you, too - each disk will
presumably be assigned to a volume group. lspv will show this VG, so you
can tell which disk goes with which system.
For the non-HA systems I've just relied on good, (well; adequate)
documentation.
SSA will, of course, survive a broken link caused by the loss of one of the
servers in the loop. However, be aware that if you have two servers out you
may end up with isolated disks. This depends on your cable layout, of
course. Because of this it is a good idea to have VGs in shared loops set
NOT to vary on at startup, even if they're not involved in your manual HA
set-up.
Also, if you ever have occasion to re-boot the servers you should do it one
at a time. If you re-boot two servers simultaneously you can have problems
when they come to configure the SSA disks and one of the servers may fail to
properly configure the disks, or even hang. In fact, if the servers are
similar then you can simply stagger the startup: leave a few minutes between
each one. That way they don't end up doing SSA configuration at the same
time.
Regarding the drives being reserved, (Holger's post), there IS a way of
clearing this without HACMP. It's a long time since I've thought about it
and I can't remember exactly what was involved, but I'm pretty sure it was
done through the SSA Service Aids, in the diagnostics, and it wasn't a
specific task to clear the reservation. If I remember more I'll let you
know.
Simon Green
Philip Morris ITSC Europe
AIX-L Archive at http://marc.theaimsgroup.com/?l=aix-l&r=1&w=2
AIX FAQ at http://www.faqs.org/faqs/aix-faq/
N.B. Unsolicited email from vendors will seldom be appreciated.
> -----Original Message-----
> From: Andrew Garnett [mailto:andrewwdg@YAHOO.CO.UK]
> Sent: 19 June 2002 17:05
> To: aix-l@Princeton.EDU
> Subject: Shared SSA Drives
>
>
> Hi All,
> We have just (for the first time) reconfigured our SSA
> drawers so that 4 systems are in a loop with all their
> SSA drawers. We intend to develop a manual failover
> process to exploit this.
>
> Of course, this means that each system now has a large
> number of "hdisks" and pdisks belonging to the others.
>
> Now, I can document these, or a process to identify
> them, but is there a recommended approach to making
> them identifiable on all systems ?
> Should I be renaming the "hdiskn" to "node1_hdisknn" ?
> What do you use ?
> And come to mention it, is there a recognised document
> for this kind of failover ?
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:16:00 EDT