Hi all,

I have an aging CentOS5 system, which acts as an internal mail (Postfix) relay and VPN server. It's ran for years very reliably, but today, it suddenly stopped sending mails out.

The mailq is growing and I can't see why.

Only two people have root access to this server (me and my colleague) and neither of us are aware of anything we've done to upset it.

Since mails stopped being sent on, the following appears in our maillog:

"
Sep 7 15:12:39 mail1 postfix/smtp[8702]: fatal: unknown service: smtp/tcp
Sep 7 15:12:40 mail1 postfix/qmgr[6772]: warning: premature end-of-input on private/smtp socket while reading input attribute name
Sep 7 15:12:40 mail1 postfix/qmgr[6772]: warning: private/smtp socket: malformed response
Sep 7 15:12:40 mail1 postfix/qmgr[6772]: warning: transport smtp failure -- see a previous warning/fatal/panic logfile record for the problem description
Sep 7 16:12:40 mail1 postfix/master[5979]: warning: process /usr/libexec/postfix/smtp pid 8701 exit status 1
Sep 7 16:12:40 mail1 postfix/master[5979]: warning: /usr/libexec/postfix/smtp: bad command startup -- throttling
"

The Postfix config is largely default. It isn't chrooted.


I have:
Restarted Postfix.
Restarted iptables (and openvpnas).
Rebooted entire server.
Checked existence of /etc/services exists - it contains the smtp service.

I am weary of making drastic changes, as, like I say, it's ran for years without incident. But am prepared to do so now!

Any suggestions would be appreciated.

Thank you!

Regards,

Elliot

OS agrees that Postfix is my default MTA:

[root@mail1 yum.repos.d]# alternatives --config mta

There are 2 programs which provide 'mta'.

Selection Command
-----------------------------------------------
* 1 /usr/sbin/sendmail.sendmail
+ 2 /usr/sbin/sendmail.postfix

I can telnet out to SMTP servers okay:

[root@mail1 postfix]# telnet aspmx.l.google.com 25
Trying 74.125.140.26...
Connected to aspmx.l.google.com (74.125.140.26).
Escape character is '^]'.
220 mx.google.com ESMTP r8si2182368wrc.0 - gsmtp
QUIT
221 2.0.0 closing connection r8si2182368wrc.0 - gsmtp
Connection closed by foreign host.

Privileges on /etc/services seem okay (Postfix should be able to read it):

[root@mail1 etc]# ls -lha services
-rw-r--r-- 1 root root 354K Sep 7 15:57 services

I think I have got to the bottom of this.

I believe a script we have running, which chmod's and chown's a directory, failed, whereby the variable holding the directory path was empty, so / had the privileges change applied!
Resetting the privileges on / has sorted it.