We recently had a failure issue with our hardware here at Georgia Tech. One of the disks on ops crashed right before we could mirror it and back it up. We had /users and /proj on that disk. So the question we have is, is it possible to recreate just this part and hope that we can get the software to work or will we have to reinstall the entire system again? If we can recreate a portion, could you please point us to what we need to do?
I do not think this has come up before, so we do not have any ready
made tools to recover from it.
However, I think things are written in a way that will let you get
going again. Making this up as I go ...
First recreate all the users. Be sure to prune out your own uid.
foreach my $uid (@alluids) {
boss> sudo pw userdel $uid
boss> wap tbacct add $uid
}
Then recreate all the projects:
foreach my $pid (@allpids) {
boss> wap mkproj -s $pid
}
The reset all the groups for all the users.
foreach my $uid (@alluids) {
boss> wap setgroups $uid
}
The above lists can be gleaned from the users and projects table in
the DB.
mysql> select uid from users;
mysql> select pid from projects;
I suggest making a backup of your DB first.
boss> mysqldump tbdb > db.backup
Let us know how this goes please. It would be nice to get a FAQ entry
into the wiki that describes the process completely.
Lbs