Ignite Restore Problem

[Lars-Olof Fermvall]

Last year we installed Ignite 1.59 as part of a number of system deliveries. The command make_recovery -vAC was used to create a backup of the root disk, which is the only member of vg00. The root disk could be successfully restored from this backup, and also the LVM configuration for other volme groups as well as fstab was correctly restored. Now the customer reports that this is no longer the case, but that the information about all logical volumes, except the root volume group, is lost when the backup tape is re-loaded from tape. Also, an fstab version without the other volume groups is loaded (The original one is saved). All that I know of that has changed is that patches have been installed - attached is the output of an swlist command.

[Bill McNAMARA_1 · Accepted solution]

That's the way ignite works.
It only backup's vg00 (when using -A)
You must backup all other filesystems the
traditional way.

[Patrick Wallek]

That's not entirely true. If /usr is not a part of vg00, but a part of vg01, then all of vg01 will get backed up as well.

Do a man make_recovery for more specific details on this behavior.

Also, I would upgrade to the latest version of Ignite. I think it is up to 2.6 now.

[Henry Weldon]

Use should use the lastest version of Ignite-UX and have the lastest "pax" patch installed too. Create your Ignite tape with make_recovery -vCA, add the -i flag if you plan to interact with the recovery. If you do not create the tape with the -i option and then you interupt the recovery and access the advanced functions, I have seen it create a min. system with bare bone files. This is usually not what you want.

[Bruce Regittko_1]

Hi,

Ignite should backup the LVM information that is present in /etc/lvmtab, etc. However, if your tape was made a year ago and any changes have been made since then, then you may have a problem.

If you used the -C option to make_recovery, you can run check_recovery periodically to see which files have changed and make a decision whether or not to re-run make_recovery.

--Bruce