cancel
Showing results for 
Search instead for 
Did you mean: 

Site-failover and Trucluster

Eric van Dijken
Trusted Contributor

Site-failover and Trucluster

While trying to create a test cluster on our other site. I keep getting the following message "Waiting for Cluster Member Boot disk to become registered"

The disks come from a ES45/MC cluster, using EVA 5000 san disks. Of these disks we made a CA replication set, to our other site. Than we made "Clones" from this set. Which we presented read/write to the Test cluster.

On other message which keeps coming up is, the one with "ccfg_MakeDeviceIdentWWID"

Booting the Cluster single-user mode, doesn't help either.

From what i know/expected this should work, without much problems....... But as it is, i am stuck.

Anyone ever done this, or any pointer on what this messages really mean?

Watch, Think and Tinker.
10 REPLIES
Vladimir Fabecic
Honored Contributor

Re: Site-failover and Trucluster

Hello Eric
Can you tell what OS version you have and what patches are installed. Do you have fibre channel tape device? I had such a problem with tape device. Try to isolate or remove FC tape device for test. From my experiance, this should work.
In vino veritas, in VMS cluster
Eric van Dijken
Trusted Contributor

Re: Site-failover and Trucluster

The OS version is 5.1B-2 (NHD7 / PK4) and should have the latest ERP patches.

Yes, we have fibertape drives attached (ESL-712, with 8 LTO drives)

What really bothers me, is that i can boot from the disks. But even in single user mode, i can't do a thing. Everything is read-only, can't even read the disklabel in (-r) mode. Tried a /sbin/dn_setup -init, that failed because the read-only modus of the disk.

Watch, Think and Tinker.
Vladimir Fabecic
Honored Contributor

Re: Site-failover and Trucluster

Hello
This looks like WWID problem to me. Did you check for possible duplicate WWIDs? Does WWIDMGR in SRM console show something strange?
In vino veritas, in VMS cluster
Eric van Dijken
Trusted Contributor

Re: Site-failover and Trucluster

The WWIDMGR settings look normal. Device only shows once, with 4 paths of which 2 are connected.

Somehow it looks like the WWID's on UNIX don't match, with what is expected.
Watch, Think and Tinker.
Han Pilmeyer
Esteemed Contributor

Re: Site-failover and Trucluster

When you said you made the clone of the CA replica's, does that imply you made a snapshot/snapclone in the EVA on the remote side?

Tru64 UNIX and TruCluster are very picky about the WWID of the devices. If you made a snapshot on the EVA and tried to boot from those, then that won't work. The snaps will have different WWID's from the original.
Eric van Dijken
Trusted Contributor

Re: Site-failover and Trucluster

That would expain it.

Is there anyway to "fix" it?
Watch, Think and Tinker.
Han Pilmeyer
Esteemed Contributor

Re: Site-failover and Trucluster

That depends on what you're trying to achieve.

You could just use the replica's as those have the same WWID (on both sides of the CA connection). Other customers have used the Rob Urban scripts to have a remote copy of their system disks available on a separate set of disks (http://www.unix-wissen.de/Tru64/clone-cluster.html)
Eric van Dijken
Trusted Contributor

Re: Site-failover and Trucluster

We have a 2 site setup. 1 production server on site A and a Test/Failover cluster on Site B.

The cluster disks are replicated (CA) from A -> B.

In order to test a few procedures, we would like to use the production data on our test cluster. We can't stop (split) the CA replication, as that would be dangerous if a failure occours while we are testing.

So we made a clone of the Site B CA disks, assigned new UDID's to them, made them R/W. Now we are trying to get the Test cluster to boot from the "clones"

This is were we are stuck.....
Watch, Think and Tinker.
Han Pilmeyer
Esteemed Contributor

Re: Site-failover and Trucluster

Do I understand correctly that you have a single server on the "production site" and a cluster on the "failover site"? So the cluster is not stretched between A and B?

You don't have to stop CA to do a CA failover. That way you could boot from the same disks on the B site. However this will only work if you have identically configured servers on both sites.

You could also boot a server on the B site (again if it's identically configured) from the storage that is still presented on the A site (assuming that the SAN zoning allows this).

If the A site is indeed a single system and not a cluster, then it may be possible to script it, so that you could use the BC's.

Perhaps we should just make an appointment to talk about this, since we happen to be based in the same country.
Eric van Dijken
Trusted Contributor

Re: Site-failover and Trucluster

@Han,

You are always welcome to come by my office and have a coffee (or two) and talk about this kind of setups/configurations. Mail me for an appointment at

But let me try to explain it again. We have a site A (Es45 local MC/Cluster; production, with disks CA's to the Site B EvA)

On site B, we have a Es45 local MC/cluster; Test.

We are trying to get the Test cluster running on a copy of the Production disks. Without interuption to the Production env.

So Cluster on Site A, stays up (no downtime allowed). But we make a clone of all the disks (from the Site B, CA disks) and thus can use the production data/configuration on our Test cluster. Atleast that was the idea.

So far, we failed.

Watch, Think and Tinker.