[prev in list] [next in list] [prev in thread] [next in thread] 

List:       c-client
Subject:    RE: Ineffective ipop3d?
From:       "Rubinstein, Dmitry" <Dimrub () icomverse ! com>
Date:       1999-08-27 16:47:53
[Download RAW message or body]

I think your question belongs to the c-client mailing list rather than to
imap list (which is for discussions regarding the protocol, not the specific
implementation).

> From: Andreas Lange [mailto:lange@student.liu.se]
> Sent: Friday, August 27, 1999 11:31 AM
> To: imap@u.washington.edu
> Subject: Ineffective ipop3d?
> 
> 
> In our system we have a mailserver (2*300 MHz Alpha running 
> Digital UNIX
> 4.0E). We used to run Qpopper and UW imapd 4.5, but upgraded 
> to UW IMAP
> 4.6-beta.
> 
> After moving from qpopper to ipop3d, our system is totally swamped by
> ipop3ds. The load sometimes goes up to as much as 60, caused by 40-50
> ipop3ds. With smaller number of processes the load never drops below 3
> anyway.
> 
> As everything went smoothly before (hardly any load at all), I suppose
> there is something wrong in the latest UW IMAP 4.6-Betas. Or have we
> missed something when compiling?
> 
> Or is simply ipop3d so much slower than qpopper? Then we'll have to
> switch back.

I'm not 100% sure that this is the reason for such high load, but ipop3d is
indeed quite slow. This is probably due to one specific function with a
colorfull name 'blat()'. I beleive this problem can be fixed relatively
painlessly (and we have rewritten this function once faced with this
problem) and then ipop3d will be almost as effective as imapd (but not
quite, since there's still the issue of replacing a signle dot on a line by
double dot).

--
Dmitry Rubinstein 

[prev in list] [next in list] [prev in thread] [next in thread]