From: Milligan, Eric T. (Eric.Milligan@ngc.com)
Date: Tue Nov 23 2004 - 12:59:20 EST
Thanks to Chris Eubank and Dr. Tom Blinn for their excellent advice.
Chris wrote:
Create a Tar header with your fstab, sysconfig, and dump and restore
commands used to create the backup prior to using dump on the tape :)
Then you can just tar xvf the header and any files required for the restore,
knowing they'll always be current.
Dr. Blinn wrote:
I can't speak to the details, but I strongly suspect that what you
propose (reverting "dump" to one that was working OK for you) is
likely to be the best and simplest solution. Be sure to test this,
but since the patches are at least in theory subsettable, you are
not likely to run into problems with your method. (I personally
have found on some occasions that using older versions of layered
components solves a problem introduced by a "bug fix" in a later
version, and in general, older versions turn out to be completely
compatible with a newer kernel and libraries; so using the "dump"
from the original V4.0F kit or the one from an older patch kit is
likely to be just fine.)
We ended up actually doing the former, but placed it on a modified version of the OS Boot CD so as not to confuse our users with tape commands.
--- Eric -----Original Message----- Hello gurus - This is a call for suggestions more than a dire need for help. Basically, we base one of our products on machines running 4.0F with PatchKit 8 installed, and as part of the disaster recovery instructions we send out to our customers we use dump & restore on the UFS filesystems. As recently as PatchKit 7, we've not had any problems running dump in single-user mode to backup the systems, booting off the OS CD, and using restore from there to recover the data. However, we're hitting an error now on restores. The error message is "Cannot malloc space for property list." From the release notes (http://h30097.www3.hp.com/docs/patch/40F/bl22/ReleaseNotes.pdf) I see that patch 1193 "fixes" that problem. That patch updates not only restore but dump as well, so our theory is that problems arise when the "virgin" version of restore on the OS CD acts on files created by the patched version of dump. Has anyone else ran into similar problems, and if so how did you work around it? Our current plan of attack is to just back out patch 1193 from our PK8 systems. That way, dump and restore would be at the "virgin" state again, as on the OS boot CD. Suggestions? Thanks all, Eric Milligan
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:50:12 EDT