I had courier-0.73.1 running happily on an SL 6.5 box until the
other day when the box broke down. Motherboard, the whole machine
had to be replaced, completely different hardware, hence I couldn't
restore the entire system from backup. So I took the opportunity to
upgrade to Centos 7. Compiled courier-0.73.2 on the new system,
copied my /etc/courier from backup, done.
Or so I thought. My log started filling with entries like
courieresmtpd: started,ip=[::ffff:91.227.208.147]
and then nothing. Mail was arriving, but it wasn't being delivered.
Vanishing into thin air, it seemed, until I found it in the spool.
Imapd was working fine and existing mailboxes were available, which
means that authlib was working fine and all paths and permissions
were fine. Nothing had changed there anyway, so they should be.
Having ripped most of my hair off, I realised that although courierd
was starting, it wasn't running. Then I tried strace and found the
reason:
[pid 30489] open("/etc/sysconfig/i18n", O_RDONLY) = -1 ENOENT (No such file or
directory)
Some bright head (probably the same who thought systemd is a good
replacement for sysvinit) decided to rename /etc/sysconfig/i18n
to /etc/locale.conf.
Solution: edit /etc/courier/courierd accordningly to source the
file from its new location.
This cost me endless hours of barking up a forest of wrong trees,
so I hope that posting it will save others the same ordeal.
BTW, if you're planning a similar upgrade, take also into account
that UIDs 500-999 which were previously assigned to users, are now
reserved for system accounts. There's good potential for weird
troubles there too.
Z
------------------------------------------------------------------------------
_______________________________________________
courier-users mailing list
courier-users@lists.sourceforge.net
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
