|Veritas can suck my balls
||[Apr. 7th, 2005|10:22 pm]
Thanks to my leasing of mojo to jgp, I have found yet another oxymoron. Intelligent disaster recovery.
The concept of IDR is simple; make a bootable CD that can load a basic OS and automagically connect to the backup server (a server that holds backups, not a secondary server) to download the most recent backup to restore onto itself. It even works, somewhat. Except for when it decides to restore onto the backup server, as opposed to onto itself, for some ungodly reason likely having to do with co-workers being in a hurry.
Of course, the permissions wouldn't allow file restoration like that, so it failed. Except somehow the registry hive managed to be restored on top of the existing registry on the backup server.
Of course, this happened several days ago, and everyone noticed the weird "failed restore" messages, but didn't realize there was one component which restored successfully. Today, we rebooted the backup server, and it comes up in 640*480*16, sans novell login, bitching about hardware changes. Shit.
Totally fucked the backup server. It's like an episode of Get Smart where Hymie the robot has smoke pouring out his ears. The OS is just scrambled beyond beleif; it's like mashing two computers into one.
Of course, this happens three days after I get a new tape library setup on the thing, and mangle its config sufficiently to have organized backups in the future.
Of course, the last backup of the backup server was a week ago.
On a more positive note, I'm still the most observant in the office; noticing immediately that no red N is a big uh oh. And the server name difference on login, that was a definite hint.
I still have the unique trait of giggling at the absurdity of what happened, just like another incident long ago where the SAN array partitions were deleted by a stupid service pack. Everyone else is still going "What the fuck happened?" and being all solemn and I'm thinking this is the funniest shit ever, looking through the backup binder to find which tape to pull, because it is so clear that nothing man-made will ever make that work again.
 Not strictly necessary these days, what with to disk backups. And as for that SAN incident, we pulled a rabbit out of our ass in the form of $300 undelete software for NCS SBD NSS partitions, and didn't end up needing to restore anything from tape.
 It's a Novell thing. Novell Cluster Services Split Brain Detection NSS... NSS... You know, I dunno what NSS stands for. It's a Novell volume type, so... Novell Storage Subsystem? Named Switch Service? Nah, that's something to do with unix authentication. Fuckit, it's bedtime.