Hi,

some questions to that:
Have you already tried logging to a remote syslogd via udp ?
Does the file you redirected stderr to reside on the same
drive/slice as your mail.* file (probably /var/log/maillog) of
syslogd? How is the slice mounted syslog is logging to?
Is your dbmail database on this drive/slice, too?
You didn't expect a system with debugging log level (i.e.
TRACE_LEVEL=5) turned on to be awesome fast, did you?

I cannot imagine that syslog is a stopper, because postfix
normaly logs to syslog, too and is awesome fast...
Looking at cyrus and uw imapd - they are normaly logging
to syslog, too. meaning: syslog is *the* logging standard in the
*NIX world.

If you change the default logging to a seperate file, (like say,
apache) you give up some very nice features - like logging to
remote hosts, databases (msyslogd) which must then be added
again by the administrator through an ugly pipe interface to
logger, if  logging to remote hosts/databases is desired.

The only thing I'd like as improvement in logging are
configureable logging-facilities so imapd could log to local3.*
pop3 to local2.* for example and self explanatory instead of
numeric log-levels

--
Wolfram

Mikhail Ramendik wrote:

Hello,

I have found the culprit that makes fetching slow.

Unbelievable, but it's none other than the syslog calls in trace() !

When I hacked it to have syslog off and verbosity to stderr on (the -v
switch does not work for some reason), and just redirected stderr to a
file, I got over 180 messages per second ! (With both of my patches - I
think they're still needed).

Therefore I'm writing to tell users that the workaround is to lower the
TRACE_LEVEL as low as you can live with.

And, to tell developers that we probably need a log file of our own. I'm
unable to implement it as I am not familiar enough with the C library,
and with the code (to understand where the file should be opened and
closed), and with working with files from forked processes.

This is a MAJOR speed issue. I think it's a blocker for 2.0.1.

Yours, Mikhail Ramendik



_______________________________________________
Dbmail-dev mailing list
Dbmail-dev@dbmail.org
http://twister.fastxs.net/mailman/listinfo/dbmail-dev

Reply via email to