VegBank

colby emailed 6/28/06 about this.

colby has responded and backups will be easily done. We should definitely shift our strategy to putting the latest copy into its own directory. Then we can move it out before adding a new one, so that we only get the latest one or two backed up on the main NCEAS server when that happens.

I have updated /vegbank/bin/vbdump in CVS to do this:
--- dir explanation ---
/usr/vegbank/backup -- initial location of backups
/usr/vegbank/oldarchive -- any archives that are not the latest wind up here
/usr/vegbank/latest -- this dir contains the latest backup(s)
/usr/vegbank/latest/to_move -- this dir contains old backups for which a replacement exists, but hasn't been copied into /usr/vegbank/latest yet.

--- what happens ---
The backup initially goes into /usr/vegbank/backup.
Then it moves everything out of latest and into latest/to_move
Then it moves the db into /latest
Then it moves everything out of latest/to_move and into oldarchive

Problems still remaining:
PMark set this up to scp. I can't get it to do this unless I supply a password (in which case it works, but that's not automation)
Postgresql doesn't allow access without a password, either. So Chad, is there a way around that for this script? If it was run as the vegbank user, perhaps we could trust that user for accessing the database if the user is really logged in?

colby has confirmed that this is working. //vegbank/usr/vegbank/backup/latest is being backed up to tape with NCEAS stuff.

Original Bugzilla ID was 2481