RE: BDB recovery after power outage

["Luke A. Kanies" <luke@madstop.com>]

On Sun, 20 Apr 2003, Howard Chu wrote:

> Hm, let me see, you already had one major outage before and you still don't
> have backups, and you clearly haven't read any of the documentation. Yes, it
> certainly seems like there's something wrong with your thought processes.

It's quite an assumption on your part that I haven't read any of the
documentation, when it's pretty obvious that one must read at least some
documentation in order to get OpenLDAP running at all.  Considering I've
currently got it running as the backend for virtual emailing and virtual
ftp domains, I must have read _something_, huh?

But I definitely thank you for being insulting; that always really adds
something to a discussion.

As to the persistent lack of backups, unfortunately, this failure came too
closely on top of the previous failure for me to have backups in place.  I
wasn't even sure that I had all of the data when I had the second failure.
This is a small system and is not my day job, so I can't exactly spend the
time it apparently requires.

But hey, you can bet I do now.  God forbit I expect a system to function
for a week without catastrophic failures.

> You might start reading here:
> http://www.openldap.org/faq/data/cache/738.html

Yeah, I've read that before, and I've read most of the documentation it
points to.  My problem isn't really that I can't figure out how to
recover, it's that I don't want to have to recover.  I could understand
having to go through a recovery procedure if someone were writing to the
database, but that wasn't the case; only two people have write access to
it currently, and neither of them were writing.

So the question really is, why is that a database not getting written to
(and one which hadn't been written to for > 12 hours) must be recovered if
it fails badly?  That seems like a pretty serious bug, especially one
which occurs in software which specifically is set up for more reading and
less writing.

And BTW, if it matters, the recovery for this was especially retarded.
Even though I couldn't see the data with an ldapsearch, I could see it
with a 'slapcat', so I used slapcat to get a copy of the data, and then
used slapadd to add it back in, with a -c to ignore existing entries.

If I had a database with lots of writing, that would probably cause
problems, but it seemed to work fine, and now I can get back to actual
work (tm).

Luke

-- 
There are three kinds of death in this world.  There's heart death,
there's brain death, and there's being off the network.  -- Guy Almes