Re: bdb backend - reliable or not?

[Atom Powers <apowers@digipen.edu>]

Tomasz Chmielewski wrote:
> Atom Powers wrote:
>
> (...)
>
> > Then put the db_recover command in your start-up script, and/or tune 
> > BDB so that it can survive a crash more easily. 
>
> Is it OK if I put "db_recover" before each slapd startup? I guess it 
> should be avoided.
> In that case, how can I detect that the server crashed unexpectedly, and 
> script "db_recover" run?
> Should I only look at existing "alock" file before starting slapd, and 
> if it's there, assume something bad happened, and run "db_recover"?

I just put it in the start-up script, to run every time slapd 
(re)started. If the database wasn't corrupt, no problems would need 
fixing, so db_recover didn't do anything.

Disclaimer: This was with an older version of OpenLDAP, I plan on having 
enough redundancy with the new system to detect and recover from any 
problems manually.

> > But if you are experiencing unexpected crash/poweroff I'd say you have 
> > more serious issues than whether to use ldbm or bdb.
>
> No, I just prefer to be safe than sorry, and do the things "right".

The "right" way is not to let your servers die. Use a reliable OS on 
reliable hardware, with sufficient power backup and locked away from 
people, including yourself, who may accidentally unplug something. ( Of 
course you still need backups and a recovery plan, but that is for 
disasters and shouldn't be required except in extrema circumstances. )

--
Perfection is just a word I use occasionally with mustard.
--Atom Powers--
Systems Administrator
DigiPen Institute of Technology
(425) 895-4443