On 10-10-2024 15:11, Christian Schulte wrote:
On 10/10/24 11:13, Mark Lawrence wrote:
On Thu Oct 10, 2024 at 10:48:24AM +0200, Christian Schulte wrote:
You can enable tracing which will log the information you are looking
for. For example:
smtpctl trace smtp
Thanks for the pointer. That is however clearly more of a live trouble-
shooting feature, which I would not want to permanently enable due to
the noise in the logs. The reason why the trace function exists in the
first place :-)
Well. You wanted to trace issues with your configuration. That's what
tracing functionality is for. You would not enable it when there are no
issues to resolve. There haven't been any issues with your
configuration. You could have easily verified that yourself by just
enabling tracing.
I don't think they wanted to trace issues with their configuration. I
think they just wanted to see what interface/port the sessionthat failed
authenticationwas established on.
My request is about connecting standard log output to the smtpd.conf
configuration in a way that it isn't connected today.
Would not add any value. For troubleshooting, just enable tracing. There
have not been any troubles with your configuration and you just got
confused. I looked at this and already had a patch around adding what
you are requesting. Does not make any sense. It's already way to noisy
the way it is in the non exceptional case. In normal operation all you
want to "hear" from a daemon is when things are failing for some reason
needing manual action, not if some connection could be established
successfully and such. Some remote connection failed authentication for
you and this got logged already. Tracing this would already have been
possible by enabling tracing.
I don't think tracing is a solution here. They would have to enable
tracing and then wait for the same thing to happen and trust it was
actually the same thing.And you don't want to keep tracing enabled all
the time, because you will get way too much spam. And currently, without
tracing, when an authentication failure happens you don't actually know
if it was one of your own users or a mail submission.
In general, having tracing doesn't make logging useless. Especially when
authentication fails it is useful to know what interface and port they
connected on (or which "listen" statement).
So to me the feature request makes sense.
-- Maarten
