Monday, August 29, 2005

why do backups have to be so big?

It's not often I talk about work, but today is a particularly unusual day. I have a pretty big development project that is due to go live next week. Everything was going pretty smoothly until some problems in testing indicated some bugs with the development platform. Just in the nick of time there was a patch released late on thursday to fix these issues, just prior to customer testing on friday. With news of the patch we pushed the customer testing back to today to give us a chance to retest under the new patched environment.
As with a lot of patches from software companies this one broke more than it fixed. But it turns out reversing the patch corrupts databases. This included my testing system, which involved applying 25 customisation sets (which take roughly 30mins to apply) to restore. So by 3am this morning I had a system at almost the point prior to patching, Yay!

1 comment:

Anonymous said...

Argh! I hate that... did a patch install on my Openview server the other week. Needless to say it broke stuff (long story). Backing it out didn't fix it. Why? After trawling through scripts for a while to figure out what it does it takes a copy of the files it's about to patch and preserves them somewhere else and then puts them back in your "backout".

Unfortunately being Unix they did a "cp" which means the copies get owned by the user who made the copy... and being a patch it was run as root. So when the copies were put back their file ownerships were all screwed up. Brilliant!

You'd think these software companies would know more about what's going on... but sadly... they don't. BEWARE!!