topic Re: Why mail loops? in Operating System - OpenVMS

Why mail loops?

Markus Waldorf_1 — Fri, 03 Jul 2009 11:24:40 GMT

Sorry, me again, and starting a new subject about a pesky smtp mail problem. I can not affort to just give it up and need to have it fixed by tomorrow night. I have a simple setup now.

Problem:

When I send smtp mail to myself using the NW1LAN network interface I receive mailing loops or uknown host, but when I use NW1PUB it works fine. Both NIC's are in the same computer.

From: SMTP%"TCPIP$SMTP@nw1pub.xx.yyy.org"
To: system@nw1pub.xx.yyy.org
CC:
Subj: Returned mail

%TCPIP-E-SMTP_EXCMAXHOP, maximum number of hops exceeded; mail loop suspected

Return-Path: system@nw1pub.xx.yyy.org
Received: from nw1lan (192.168.100.100)
by nw1pub.xx.yyy.org (V5.1-15Q, OpenVMS V7.2-1 Alpha);
Fri, 3 Jul 2009 13:09:05 +0200
Received: from nw1lan (192.168.100.100)
by nw1pub.xx.yyy.org (V5.1-15Q, OpenVMS V7.2-1 Alpha);
Fri, 3 Jul 2009 13:09:03 +0200

--> When I turn SMTP relay off, it shows

From: SMTP%"TCPIP$SMTP@nw1pub.xx.yyy.org"
To: system@nw1pub.xx.yyy.org
CC:
Subj: Returned mail

%TCPIP-E-SMTP_UNKHST, remote host unknown, nw1.xx.yyy.org
-SYSTEM-F-NOSUCHNODE, remote node is unknown

Date: Fri, 3 Jul 2009 14:17:01 +0200
Message-Id: <09070314170128@nw1pub.xx.yyy.org>
From: system@nw1pub.xx.yyy.org
To: SYSTEM@nw1.xx.yyy.org
Subject: tset
X-VMS-To: SYSTEM@NW1.xx.yyy.org

SYSTEM@NW1>ucx sho int
IE1 192.168.100.100 255.255.255.0 934 594 1500
LO0 127.0.0.1 255.0.0.0 1173 1173 4096
WE1 172.18.152.128 255.255.255.0 4841 2644 1500

SYSTEM@NW1>ucx sho hosts /local

LOCAL database

Host address Host name

127.0.0.1 LOCALHOST, localhost
192.168.100.101 dns1
192.168.100.201 dns2
192.168.100.100 nw1lan, NW1LAN
172.18.152.128 nw1pub, NW1PUB
192.168.100.200 nw2lan, NW2LAN
172.18.152.129 nw2pub, NW2PUB
172.18.152.1 nwgwpub

SYSTEM@NW1>ucx sho conf smtp

SMTP Configuration
Options
Initial interval: 0 00:30:00.00 Address_max: 16 EIGHT_BIT
Retry interval: 0 01:00:00.00 Hop_count_max: 16 RELAY
Maximum interval: 3 00:00:00.00 NOHEADERS

Timeout Initial Mail Receipt Data Terminate
Send: 5 5 5 3 10
Receive: 5

Alternate gateway: not defined
General gateway: not defined

Substitute domain: not defined
Zone: not defined

Postmaster: TCPIP$SMTP
Log file: SYS$SPECIFIC:[TCPIP$SMTP]TCPIP$SMTP_LOGFILE.LOG

Generic queue Queues Participating nodes

TCPIP$SMTP_NW1_00 1 NW1

SYSTEM@NW1>ucx sho name

BIND Resolver Parameters

Local domain: xx.yyy.org

System

State: Started, Enabled

Transport: UDP
Domain: xx.yyy.org
Retry: 1
Timeout: 4
Servers: dns2, dns1
Path: No values defined

Process

State: Enabled

Transport:
Domain:
Retry:
Timeout:
Servers:
Path:
SYSTEM@NW1>

Re: Why mail loops?

Markus Waldorf_1 — Fri, 03 Jul 2009 11:42:44 GMT

When I connect to nw1lan
the SMTP answers with nw1pub

SYSTEM@NW1>telnet nw1lan.xx.yyy.org /port=25
%TELNET-I-TRYING, Trying ... 192.168.100.100
%TELNET-I-SESSION, Session 01, host nw1lan.xx.yyy.org, port 25
220 nw1pub.xx.yyy.org V5.1-15Q, OpenVMS V7.2-1 Alpha ready at Fri, 3 Jul 2009 14:40:37 +0200

why does it choose nw1pub?

Re: Why mail loops?

Steve Reece_3 — Sat, 04 Jul 2009 00:23:09 GMT

I don't see any details for an alternate gateway or for the routing table on the system.

What was the address that the mail was meant to be getting sent to (i.e. the outgoing message that resulted in the errors/loops)?

nw1 or nw1.xx.yy.org isn't in the host table - does the system really know what/where/who it is in IP terms, as opposed to a DECnet cluster alias?

SMTP needs to know who and where nw1 is, whether it's an alias for the local host or it's some other host elsewhere. If it's getting the address that's on the other NIC then it may be trying to go out onto the network, come back in the other interface, and getting itself confused. nw1 needs to be either in the host table or resolvable by TCP/IP by DNS lookup.

The alternate gateway is there in the SMTP configuration to tell SMTP where to send mail that isn't for the local network. If you're using an Exchange server or a Unix/Linux mail server to process outgoing mail and that's where this mail is meant to be going then that server's address needs to be in the Alternate Gateway field.

Either way, you may want to include a substitute domain so that the mail that comes back as a reply to the message that you're sending out goes to the correct place and the correct mailbox.

What does the SMTP log file have to say?

Steve

Re: Why mail loops?

Steve Reece_3 — Sat, 04 Jul 2009 00:25:51 GMT

The output of $ SHOW NETWORK might be useful too and will probably tell you why the machine is responding as nw1pub when you expect it to be responding as nw1lan

Re: Why mail loops?

Markus Waldorf_1 — Sat, 04 Jul 2009 15:45:05 GMT

Thanks a lot for taking a look at my problem! Unfortunately I don't have the same setup anymore. I got it working to some stage but I don't think it's a solution.

I did one more test, reconfigured the 172 interfaces from scratch. The tcpip$config did all the host entries. I did not setup the 192 interfaces yet, but configured the local bind resolver to use it. The result is each host can send mail to itself, but sending mail from host1 to host2 takes about 20 min. I do not use cluster alias and the host file is shared by TCP/IP inside the cluster. It has only the correct entries for the 172 interfaces. When I configured the 192 interface to have traffic to the DNS server, loops and the usual problems were back.

What I have found out is that the SMTP listener appears to register with the interface that gets setup first. I worked around the problem above by cleaning out all tcpip settings again, than installing the 192 inteface first, setting up SMTP and than the 172.

Now I can at least send mail between the hosts and to another computer connected on the 192 interface (TCP/IP 5.3 and BIND), but MX using DNS MX records either result in host unknown, or DNS a-record loop, host pointing to itself. Although nslookup shows the correct names and ip addresses, as well as $tcpip sho mx yy.xxx.org. I turned on level 5 logging. The log files shows no problems, it finds the MX records, picks the correct one, talks to the host and delivers the mail. However, it comes back saying no such user system.

I will start another thread an will try to give more details about the current setup.

Thanks,
Markus

Re: Why mail loops?

marsh_1 — Sat, 04 Jul 2009 16:28:56 GMT

hi,

could you list up your perm and dynamic routing table ?

Re: Why mail loops?

Markus Waldorf_1 — Sat, 04 Jul 2009 18:37:46 GMT

Hi,

certainly. I have only a static default route defined for the interfaces in the 172 network. For the 192 network I don't need one, it would infact be itself, but I can only define one default gw in UCX anyway.

SYSTEM@NG2>ucx sho route /perm
PN 0.0.0.0 172.18.136.1

SYSTEM@NG2>ucx sho route
AN 0.0.0.0 172.18.136.1
AH 127.0.0.1 127.0.0.1
AN 172.18.136.0/24 172.18.136.65
AH 172.18.136.65 172.18.136.65
AN 192.168.100.0/24 192.168.100.201
AH 192.168.100.201 192.168.100.201
SYSTEM@NG2>

SYSTEM@NW2>ucx sho route /perm
PN 0.0.0.0 172.18.152.1

SYSTEM@NW2>ucx sho route
AN 0.0.0.0 172.18.152.1
AH 127.0.0.1 127.0.0.1
AN 172.18.152.0/24 172.18.152.129
AH 172.18.152.129 172.18.152.129
AN 192.168.100.0/24 192.168.100.200
AH 192.168.100.200 192.168.100.200

Btw, I started a new thread....