Hello.
> Stability would be an issue as well. If the LDAP server crashes or
> restarts or whatever your mail breaks instantanously as well. This is a
> bad thing (tm).
That's a separate issue. It can be addressed by running a caching LDAP server
on each machine and ensuring that they're up before trying to start exims.
> Besides, some people don't use LDAP and would have to learn just to make
> Exim work. It adds another layer of complexity that no one needs.
And I'm not suggesting an unconditional move to LDAP-based config as well.
I'm evaluating whether writing a patch and testing will worth it.
> Ah, but the configuration file is read only once! All the processes
> that are forked from the deamon (you _are_ running it in deamon mode,
> yes?) inherite all the configuration options from the deamon.
All configuration is read from configure file and stored either in global
variables or some lists. The fork() takes care of them.
> If you are really worried about changing the configuration and it
> working as by magic, add a small statement that stats the confige file
> and if it's newer than the deamon process then HUP the deamon. Add this
> just before the deamon calls fork(). Nice and easy but wastefull as
> it's even more disk IO which you do not need!
Won't cut it. Right now I have no problems with HUPping daemon whenever I
change the config file, but I'm quite tired of writing super-intelligent
perl scripts to parse it and automagically put needed features into it.
This also leads to storing the config data in multiple places, and running
a process via ssh to edit it, which proved to be quite slow and error-prone
if done on regular and automated basis.
> It is good practice to test things on a test machine and then roll it
> out to production systems. Yes, it's hassle. Yes, it will save you
> time and stress.
