Telnet connection refused by IPv6 loopback

[rawlingsta]

Working on the Administrator test. I have postfix set up and have verified init_interfaces and protocols are both set to 'all'. IPv4 is working fine. I'm able to telnet to port 25 just fine. However, when I try to telnet to IPv6 (even to ::1), I get connection refused. I'm racking my brain on this and cannot figure out what else to troubleshoot :( Any help pointing me in the right direction is very much appreciated.

[jimb]

I did it with postfix too but it was such a long time ago I can't remember how I did it.  Check ip6tables and such to make sure it's not blocking port 25.

[rawlingsta]

Good idea! But alas

Code Select Expand

ip6tables -L
Chain INPUT (policy ACCEPT)
target     prot opt source               destination

Chain FORWARD (policy ACCEPT)
target     prot opt source               destination

Chain OUTPUT (policy ACCEPT)


[rawlingsta]

Looked at the logs and kept seeing:

Code Select Expand

Jan 24 19:22:43 hostname postfix/postfix-script[6979]: refreshing the Postfix mail system
Jan 24 19:22:43 hostname postfix/master[6616]: reload configuration /etc/postfix
Jan 24 19:22:43 hostname postfix/master[6616]: warning: ignoring inet_protocols change
Jan 24 19:22:43 hostname postfix/master[6616]: warning: to change inet_protocols, stop and start Postfix


So, did a postconf and see inet_protocols is set to ipv4 where I explicitly added inet_protocols to main.cf, but won't be accepted. I feel like I'm getting slightly closer to figuring this out....

[rawlingsta]

LOL, I could learn to read. I kept restarting the service versus stopping and then starting... Sorry for the wasted thread!

[marcusw]

LOL, I could learn to read. I kept restarting the service versus stopping and then starting... Sorry for the wasted thread!

That's weird...I thought that they were the same? Last I looked at the insides of an /etc/init.d, the restart function was just about precisely the contents or stop and start back to back. Strange.

[rawlingsta]

I would have thought the same thing. Though I didn't use /etc/init.d/postfix start|stop|restart . I just used $postfix reload. Must not be the same.

[jimb]

That probably just HUPs the server daemon or something.

[marcusw]

I would have thought the same thing. Though I didn't use /etc/init.d/postfix start|stop|restart . I just used $postfix reload. Must not be the same.

Oh yeah, that does something different. Probably reloading the configs?

[thixotropy]

I would have thought the same thing. Though I didn't use /etc/init.d/postfix start|stop|restart . I just used $postfix reload. Must not be the same.

Oh yeah, that does something different. Probably reloading the configs?

"postfix reload" tells the master process 1) to reread its config,
2)tell the running processes to exit (at their leisure, so that what they're doing gets finished)
3) start new versions as they're needed, with the new configurations
The master process continues to run.  There should be no interruption of service.

Stop stops all processes.  Start starts them.  There's a period of time (short, usually, but on a busy server it can be a fairly long time) between when processes start shutting down and when they finish.  There's another interval while the master starts up.  So there's an interruption of service. 

There are a very small handful of things that the difference matters to.  Upgrading the binaries is the obvious one, but changing the addresses (inet_interfaces and inet_protocols) that postfix is listening to is one of them.  (The only others I can think of are changing the queue_directory and daemon_directory, but I don't think anyone does that often....  There might be a handful of others.  The log file should tell you about them.  )