auto-dbm problems

On Wed, 13 Dec 2000, Gert Doering, Netmaster wrote:

> On Wed, Dec 13, 2000 at 04:58:21PM +0100, Marcus Rist wrote:
> > About 5 hours seems to be "normal" right now.
> 
> Not that this is really satisfying - at good times, auto-dbm response
> has been nearly instantaneous (2 minutes).
> 
> I'm convinced that the new database will fix this, once and for good.

Gert,

The latest delay was not due to the database itself, but rather with PGP.  
We currently use the commercial PGP offering, and when you run "pgpv" on
an encrypted file, it blocks waiting on TTY input, even when you use the
--batch option.  It works properly with *signed* messages, just not with
encrypted messages.  Unfortunately, this pauses the queue for other
updates.

We do try to monitor the queue length to check for this.  It doesn't
happen often (maybe once a month), but when it does a delay can occur.  
I'm DBM this week, so any delays yesterday were my fault.

In the new implementation, we intend to use GnuPG, which does not have
this problem.

--
Shane Kerr <shane at ripe.net>
Database Software Engineer
RIPE NCC
+31 20 535 4427