Conference ssag::ask_ssag

6639.0. "raid 410 on Prioris ZX 6200" by VIVIAN::P_GRIFFIN () Thu May 01 1997 08:51

Hi Cluster-Busters

We have a new installation of two Prioris ZX 6200 connected to a RAID410.
The RAID410 has two groups of drives, failover on W-NT is set up such that one
Prioris is primarily responsible for one group of drives, and the other PRIORIS
is responsible for the second group of drives.

When we boot the first PRIORIS, it shows it has picked up its allocated group
correctly (ie GREEN state), and the other group 'failed-over' (ie YELLOW state).
When we boot the second PRIORIS, it takes over its allocated group correctly 
(ie GREEN state), and the first PRIORIS drops its responsibility for this
    group to RED state.
This is how we expect things to happen.

When we shut down either system, the remaining system adopts the failed-over
group, again we are happy bunnies at this point.

BUT, when we bring this system back up, it doesn't take over its responsibility
for its disk group. Further investigation shows that it doesn't see the group
(ie CONTROL PANEL then SCSI ADAPTER doesn't see a drive). When we use the
ADAPTEC ^A Utility, the presence of the group is random, but now we see from
Note 5787 that this is acceptable. I assume the higher layers of software can
work around whatever problem it is the Utility has to detect the group's
presence; but it seems that this would be a nice explanation of our problem.

We have changed all host to raid SCSI cables and terms, and are about to P1
a RA410 Controller. If that fails I guess I'm into IPMT land. But I would like
someone to verify that they have seen this failover scenario work on a PRIORIS
ZX 6200 configuration (it wouldn't be the first time I've seen a faster system
break a software timing loop....).

Thanks Pete

Have you asked in the LJSRV1::WINNT-CLUSTERS
conference?

    Yes, we've tested the Prioris 6200ZX systems in our NT Clusters lab. 
    But probably only with a RAID450 storage box.
    
    Did you check the NT registry as suggested in one of my replies to note
    5787?  That will tell you if NT's device scan found the disks during
    boot.  Also check the HSZ firmware rev and HSZDISK.SYS versions as
    suggested in the previous note.

    Whoops,
    Thanks for the quick help, sorry to have wasted your time, but....
    We swapped the AHA2944W (KZPEA-BY) on the first PRIORIS (we had already
    tried this on the second PRIORIS) last night, and it now appears to
    failover and fallback every-which-way okay.
    Regards Pete