Too Many Hops...

[ScatterBrain]

Can someone explain this to me please;

Code: Select all

May 10 08:38:39 concorde sendmail[18476]: k4ACcd1w018476: SYSERR(root): Too many hops 26 (25 max): from <root@firebrand.nei-ky.com> via localhost.localdomain, to <sysadmins@nei-ky.com>

That's what my Scalix machine gets when I send a message from one of my servers.

Here's the path the message should take:

Firerband ==> Etch ==> Concorde

Firebrand is the server in question. It uses Postfix to handle e-mails.

Etch is my Spam/Virus Filtering Gateway. It uses postfix, amavisd-new, spamamassassin and clamav.

Concorde is the Scalix server.

Both Etch and Concorde on inside my firewall, Firebrand is directly on the Internet at a colo site. As best as I can tell from the logs on the three machines, the message takes less than 10 seconds to get from Firebrand to Concorde. While I don't know how many routers are on my ISP network, I would think that there should only be a couple between my LAN and the Colo site (It's at the ISP's facility).

I did a tracepath from the colo box to my firewall and it took a total of 4 hops to get there:

Code: Select all

root@firebrand:~# tracepath victory.nei-ky.com
 1:  firebrand.nei-ky.com (64.191.130.98)                   0.136ms pmtu 1500
 1:  64-191-130-97.service.qx.net (64.191.130.97)           0.807ms
 2:  64.191.129.25 (64.191.129.25)                          0.899ms
 3:  18th.qx.net (208.235.88.4)                             1.234ms
 4:  no reply


Once at my firewall, it should have jumped straight to Etch, which would have then dumped straight to Concorde:

Code: Select all

etch:/var/log# traceroute concorde.nesbitt.local
traceroute to concorde.nesbitt.local (10.200.8.249), 30 hops max, 40 byte packets
 1  concorde.nesbitt.local (10.200.8.249)  0.226 ms  0.214 ms  0.163 ms

So where are the other 21 hops? Why did this message fail?

*confused*

[ScalixSupport]

Hi,

you are confusing IP routing with mail routing -:-)

First, check on concorde if /etc/mail/local-host/names has your new domain name. If not, add it.

Cheers,

Sascha.

[ScatterBrain]

you are confusing IP routing with mail routing -:-)

OK, so is this then true?

Firebrand ==> Etch ==> Concorde = 3 hops? If so, it's still a far cry from the 25 allowed.

First, check on concorde if /etc/mail/local-host/names has your new domain name. If not, add it.

Contents of the /etc/mail/local-host-names file:

Code: Select all

# local-host-names - include all aliases for your machine here.
concorde.nesbittengineering.com
concorde.nei-ky.com


Both of those names are correct and proper.

[ScalixSupport]

The local-host-names should read

nei-ky-com
nesbittengineering.com

These are aliases for concorde.

The email address for sysadmins is not configured in scalix. If it was the message is would have been accepted. Since concord didn't know that it was also nei-ky.com then it was rejecting the message back to firebrand. Firebrand knows that nei-ky.com is concord. Round and round she goes. If concord new it was nei-ky.com and the email wasn't configured the sender would have received a rejection message.

Thanks,
Don