??s on booting from DAT

[snuttall]

We have a dual server HP k class 9000 system with clustering enabled.

Last night while performing a backup of server1, I encountered an error that stated that the backup disk was not available and the backup could not continue (LVM doesn't match physical devices).

I was making a backup because our vendor didn't provide us with backups with installed software and patches. The backups we have are for previous versions of the software minus the latest patches.

I shutdown the system, replaced the bad disk 10/0.5.0 and rebooted the server.

Now, I get an error that "boot device not ready retrying boot device" on server1. It seems that the primary disk has no data to boot the system with.

Now that server 1 cannot start, the cluster will not initialize from server0. Server0 boots fine and attempts to start the cluster, but fails because the second server is down.

If I run cmruncl -n , it returns a succesful response, but cmviewcl -v indicates that package "package" is not running. In fact that package shows halted.

My questions,


1. Can I restore the bootable data on the pri disk in server1 in order to bring the system online with an old DAT?

2. Can I boot from an old DAT just enough to start the cluster and regain some order of redundancy?

3. Is it possible to extract the latest patches from the running server to patch the down server when I am able to boot it with the old software?

4. Can I force the cluster to start with only one running server?

[Steven E. Protter]

Shalom,

My thoughts by number.

1. To restore make_tape_recovery to bootable it is normally necessary to restore the entire tape.

2. No.

3. If the new server has a copy of the depot used to install the patches. Otherwise you should build a new depot and patch both systems. Extracting from /var is probably not supported and I surely don't know how to do it.

4. Depends on cluster configuration. if configured correctly with access to shared storage it should keep running and/or run the cluster. The configuration was apparently not tested prior to production, at least not adequately. A good test pre-production is to run the whole cluster on both nodes and test crashing one node to see that all cluster apps keep running.

Since it was not adequately configured you will have to look at the cluster logs after a start attempt, cmruncl or cmrunnode and correct the errors after the attempt.

SEP

[snuttall]

Shalom
Salaam
Peace

Thank you for your helpful responses. I've not been able to succesfully boot off of the tape. My concern is that the five year old backup tape is no good.

I am trying to have our Support CD located so that I can restore the boot area to the primary boot disk.

Thanks again,
SPN

[snuttall]

I have succesfully restored from tape backup.

Further issues and questions with this system are posted in this thread.

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=1103662

Thanks for your help,
SPN