2 node cluster DR test

[Timothy Weinand]

I am attempting to simulate a DR scenario with a 2 node cluster. I removed the hosts from my working storage group, created a new storage group with a couple of disks and added the hosts to it. I have a local disk so I was able to boot to Tru64. I saw the disks, labelled them and partitioned them. I restored each node root, the cluster root, /usr and /var partitions from tape. However, I am unable to boot from the console. When I try to boot, I get the error "failed to open dga100.1001.0.11.0". I no longer see that drive and my bootdef_dev variable is blank.

Anybody know what's going on?

[Steven Schweda]

> [...] I have a local disk [...]

So there are some non-local disks involved?

> [...] I was able to boot to Tru64.

On what? What are these systems?

> I saw the disks, [...]

How, exactly? What, exactly, did you see?

As usual, showing actual commands with their
actual output can be more helpful than vague
descriptions and interpretations.

> When I try to boot, [...]

How, exactly?

> I get the error "failed to open
> dga100.1001.0.11.0".

>>>show device

> [...] I no longer see that drive and my
> bootdef_dev variable is blank.

Did you set bootdef_dev before?

Did you ever see this disk from the console,
or only from Tru64? It's nice that Tru64
can find this device, but booting from it
requires that the console be able to find it,
too, and the console may be less clever than
Tru64.

Did you do any WWIDMGR stuff from the
console? (My equipment is too stupid to know
what WWIDMGR is, so I know nothing, but my
casual reading suggests that it has some
relevance to non-local disks.)

[Timothy Weinand]

Thanks for the reply.

For more information, there is a local disk that has a Tru64 installation. This was the base disk I used to originally create the cluster. The non-local disks are on an EMC SAN. When I boot with the local disk, I can see the non-local disks in hwmgr and I can label them and partition them.

I can see the devices at the console after executing wwidmgr -quickset. The issue is that I don't see a dga100 device at the console and I don't know why it's looking for it.

In doing more research, I've found out that I will need to do more than just restore the data as there are cnx partitions on the system disks that are not restored via vrestore.

[Rob Leadbeater · Accepted solution]

Hi,

As Steven requested, can you post the actual screen output that you're getting ?

If you've been removing and adding SAN storage devices, it's possible that the WWID stuff has got confused.

Probably best to clear everything out before trying another quickset.

P00>>> wwidmgr -clear all

Another thing to check is that you've got boot_reset set to ON, which is mandatory for clustered systems.

Cheers,

Rob

[Timothy Weinand]

Thanks for the replies everyone. It was determined that it's best to rebuild the cluster from the local disk. That's the plan I intend to use when this is deployed for real.

[Timothy Weinand]

Rebuilding cluster from scratch