A some-time client of mine has a mail/web server hosted at a hosting company. This server is running Debian - fairly stock config - and exim4. I was tasked, some time ago, with configuring this environment appropriately for their webapp - a primary function of which is sending mail.
Well, being unfamiliar with Debian at the time, I did my best, hacked the changes necessary into their existing config and got it working. Some time later, when they wanted SMTP authentication, I did some reading up, and found out the following:
You don't edit the config files yourself on Debian (as this company had been doing). You choose to use the monolithic config, or the split config, and then you set the relevent options using pkg-config (or add the variables into a file by hand), edit the template files, and run update-exim4.conf which will amalgamate all of these changes into the main config file, stored somewhere under /var.
So, I took this opportunity to configure the mail setup properly, using the split config. Took me a few hours one evening to get it all sorted properly. Then, thinking maybe they might see that exim4.conf was missing and think everything was broken, I put the following file, named README.IMPORTANT in the /etc/exim4.conf directory:
IMPORTANT
exim4.conf is no longer used. The debian split config scheme is now being used.
Replacing exim4.conf will cause the mail system to break.
Regards,
Wolf
Their external firewall was set to forward from a high port, externally, to port 25 on the machine, so that SMTP was accessible externally but on a non-standard port.
Everything was rocking along fine.
So, yesterday I get a mail stating that the mail system is broken, and can I please investigate. A cursory investigation shows me that everything is working on the machine, but the external port is closed. I send a mail back to this effect.
This morning, I notice that mail is working again, and think nothing much more of it.
Some time later, I get ANOTHER mail, reporting that mail is again broken.
So, apparently, their tech has checked the machine in response to a support request, found it not listening on the high port (which it shouldn't, it listens on 25), surmised therefore that everything is broken (which it isn't) and has put back a copy of exim4.conf, overriding all my changes.
*HEADDESK* *HEADDESK*
So now I have to wait for him to finish dicking about before I can check just how badly it's broken and fix it. *sigh*
ETA: I should mention that they put the port forward in place at our request, so they bloody well should have known about it.
Well, being unfamiliar with Debian at the time, I did my best, hacked the changes necessary into their existing config and got it working. Some time later, when they wanted SMTP authentication, I did some reading up, and found out the following:
You don't edit the config files yourself on Debian (as this company had been doing). You choose to use the monolithic config, or the split config, and then you set the relevent options using pkg-config (or add the variables into a file by hand), edit the template files, and run update-exim4.conf which will amalgamate all of these changes into the main config file, stored somewhere under /var.
So, I took this opportunity to configure the mail setup properly, using the split config. Took me a few hours one evening to get it all sorted properly. Then, thinking maybe they might see that exim4.conf was missing and think everything was broken, I put the following file, named README.IMPORTANT in the /etc/exim4.conf directory:
IMPORTANT
exim4.conf is no longer used. The debian split config scheme is now being used.
Replacing exim4.conf will cause the mail system to break.
Regards,
Wolf
Their external firewall was set to forward from a high port, externally, to port 25 on the machine, so that SMTP was accessible externally but on a non-standard port.
Everything was rocking along fine.
So, yesterday I get a mail stating that the mail system is broken, and can I please investigate. A cursory investigation shows me that everything is working on the machine, but the external port is closed. I send a mail back to this effect.
This morning, I notice that mail is working again, and think nothing much more of it.
Some time later, I get ANOTHER mail, reporting that mail is again broken.
So, apparently, their tech has checked the machine in response to a support request, found it not listening on the high port (which it shouldn't, it listens on 25), surmised therefore that everything is broken (which it isn't) and has put back a copy of exim4.conf, overriding all my changes.
*HEADDESK* *HEADDESK*
So now I have to wait for him to finish dicking about before I can check just how badly it's broken and fix it. *sigh*
ETA: I should mention that they put the port forward in place at our request, so they bloody well should have known about it.
no subject
Date: 2007-06-15 04:47 pm (UTC)'exim4.conf HAS BEEN REMOVED INTENTIONALLY'
.. ?
:D
no subject
Date: 2007-06-15 05:16 pm (UTC)no subject
Date: 2007-06-16 02:07 am (UTC)no subject
Date: 2007-06-15 07:55 pm (UTC)no subject
Date: 2007-06-15 08:23 pm (UTC)no subject
Date: 2007-06-15 08:35 pm (UTC)If its working there is no reason to change, but if you get a chance play with postfix sometime. I find it to be far simpler to configure.
no subject
Date: 2007-06-15 09:08 pm (UTC)Initially I went with it because ... well, yeah, it was there. And having seen enough people try to use, say, SuSE when they're from a Redhat background, and complain that nothing works and everything is broken, when in reality they're just not used to how you're expected to do things with SuSE ... I've learned to be more flexible, and adapt to the tools at hand rather than try and force them to adapt to me.
The exception to this rule is MySQL of course, which gets shown the door in favour of Postgres wherever possible. Not in this case, unfortunately.
no subject
Date: 2007-06-16 01:39 am (UTC)My thing with it too is that it's fine to configure the basics, but it would have been fairly difficult to use it as a gateway for a bunch of Exchange servers (no bridgehead for the Exchange org) the way I'm doing now.
I hope you get it fixed and running again! If something's working, then don't screw with it. I wish I had a branding iron so that I could firmly etch that principle into certain cowboy techies' brains...