Fiber Attached EMC disks on AIX 4.3.3.10

[millsra]

Here is my situation. I have 2 IBM 640's, one that is a Production Application server and One that is a DR server. We recently did a test of the DR failover using SRDF. When the server was brought up in our DR site one of the Root volumes were missing.
We also have 2 s7a's one Data base server and one DR server. We had the same problem. The 640's boot over fiber attached disks to an EMC SYMM 5 and the S7a's are scsi attached.

One of our servers has a mirrored rootvg and the other does not.
One of the volumes does not have the same target ID and LUN at the DR site as here in production.

Here is my question.
Does the ODM record target id and Lun information or does it just record the PVID? There seems to be a difference in opinion here and just wanted more input.

Oh and for those of you that are courious. We lost hd3 (tmp) hd6 (paging) and hd7 (dump) and we had 5 oracle databases up and running. Gotta love AIX.

Thanks in advance.

 

[Yegolev]

Sounds like your rootvg is on EMC disks? That does not sound like a good idea. We do have a pagingvg on EMC on many machines but rootvg is always local. IBM Certified -- AIX 4.3 Obfuscation

 

[millsra]

Yes Rootvg is on EMC disks and we have not had any problems in the past. The only problem we encountered is when we do a fail over to Philly on these 2 servers. We have not seen an issue on any other servers in the enterprise. Do you happen to know if the target id and lun are written to the ODM? If so does it use that to know where to boot. I know that the PVID is written to the odm but the target id and lun we have nto been able to get a good answer on.

 

[Yegolev]

a bit outside my experience, although from what I am led to believe all you get is the PVID. the mirroring is also odd to me, you should not have to mirror EMC disks.

we use srdf for dr but so far i have managed to avoid participating in the dr tests. IBM Certified -- AIX 4.3 Obfuscation