sendmail conection issue

[someone_4]

Hello everyone
I have a sendmail server. This is a production mail server where we are hosting email service for street end customers. We are getting complains that some customers are getting SMTP time outs to our mailserver. When I check the mailq I see allot of messages that saying with diffent from and to emails of course:
KAA27194 736 Mon Dec 3 10:37
8BITMIME (Deferred: Connection timed out with vm230.akh-wien.ac.at.)


How would I go about trouble shooting this?

Richard

[Uday_S_Ankolekar]

Hi,

Check the /etc/resolv.conf file and see the domain and name server are configured correctly at "street end" customer.
Also have a look at nsswitch.conf file for host resoultion method for your e-mail server.

Try running traceroute on both the server and test the "destination trace path"
Also try running sendmail -q 0 on customer server to flush all the queued mail

Goodluck,
-USA..

[Sridhar Bhaskarla]

Hi Richard,

Did you check with them to see if they are blocking sendmail port on their firewall?. Also, you need to check your firewalls to see if you are blocking any of their IP addresses.

This also happens if the DNS is not propagated properly like you are not able to resolve their domains from your system while they are able to see yours.

Pick up some of the messages and see if you are to connect to these systems at the port 25.

-Sri

[Tim Ryan]

If I understand this correctly, the fact that the email is being sent to the domain "akh-wien.ac.at" and is being directed to the mail or relay host vm230.akh-wien.ac.at shows that your relaying is working.

You could trying telnetting to port 25 of that particular server (vm230.akh-wien.ac.at) from your system and seeing if it is active. If it times out or you are unable to resolve the IP addresss, it could be pointing to a firewall or a remote system/network problem.

-Tim

[Steven Gillard_2]

The queued message you show indicates that your system has not been able to connect to a remote mail server. These are normal on a busy mail relay, but if you're concerned you can try telnetting to port 25 on the remote host and see if you get a response.

If your customers are complaining that they can't connect to your mail server intermittently, then you may have a local network or performance problem.

One classic sendmail gotcha is the QueueLA and RefuseLA configurations. If your 5-minute load average exceeds these configurations in /etc/mail/sendmail.cf, your server will stop delivering messages (when it exceeds QueueLA), and later stop accepting messages (when it exceeds RefuseLA). Because sendmail enjoys spawning lots of processes, the load average can become quite high, so you may want to increase these parameters if they are still at their default values.

What are the system performance metrics like? (cpu, memory, disk - run glance or "sar -u", "vmstat" and "sar -d")

Regards,
Steve

[someone_4]

here is a domain I am having problems with

# nslookup earthlinks.net
Using /etc/hosts on: lvapp13

looking up FILES
Trying DNS
Non-authoritative answer:
Name: earthlinks.net
Address: 64.225.154.175

# ping 64.255.154.175
PING 64.255.154.175: 64 byte packets

----64.255.154.175 PING Statistics----
3 packets transmitted, 0 packets received, 100% packet loss

I see that DNS is working and I get back an IP adress. But we cant ping it. So this would point to a network problem some where. Right?

[someone_4]

From what I can see the resouces are fine.

(lvapp13) /var/spool/mqueue# sar -u 1 10

HP-UX lvapp13 B.11.00 A 9000/800 12/03/01

11:29:52 %usr %sys %wio %idle
11:29:53 3 1 1 95
11:29:54 3 6 5 86
11:29:55 11 9 13 67
11:29:56 5 4 16 75
(lvapp13) /var/spool/mqueue# sar -d 1 10

HP-UX lvapp13 B.11.00 A 9000/800 12/03/01

11:30:04 device %busy avque r+w/s blks/s avwait avserv
11:30:05 c1t2d0 4.95 0.50 5 26 5.56 9.94
c1t0d0 3.96 0.50 7 79 3.71 8.55
11:30:06 c1t2d0 46.00 6.00 100 1378 26.02 24.22
c1t0d0 30.00 3.23 78 1136 11.18 14.17
11:30:07 c1t2d0 14.00 0.50 26 298 4.26 5.55
c1t0d0 4.00 0.50 8 80 5.63 4.81
11:30:08 c1t2d0 2.02 0.50 5 83 0.56 16.20
c1t0d0 2.02 0.50 1 8 6.73 14.21
11:30:09 c1t2d0 2.00 0.50 2 32 0.16 15.41
11:30:10 c1t2d0 2.97 0.50 4 38 0.29 13.42
11:30:11 c1t2d0 13.00 0.50 29 274 4.71 5.56
c1t0d0 10.00 0.50 19 200 4.72 5.69
11:30:12 c1t2d0 1.00 0.50 4 26 2.56 5.90
11:30:13 c1t2d0 3.00 0.50 5 54 1.90 13.92
c1t0d0 1.00 0.50 1 16 0.00 0.00
11:30:14 c1t2d0 11.11 0.50 25 129 5.35 4.75
c1t0d0 4.04 0.50 12 129 5.19 5.20

Average c1t2d0 10.00 3.18 20 234 14.80 15.39
Average c1t0d0 5.50 2.19 13 165 8.74 11.02
(lvapp13) /var/spool/mqueue# vmstat
procs memory page faults cpu
r b w avm free re at pi po fr de sr in sy cs us sy id
0 0 0 3455 12977 108 16 0 2 31 0 0 612 3249 135 4 3 93

[Tim Ryan]

It could be a network issue, but I doubt it. Many ISPs disable "ping" at their firewall.

I just tried pinging earthlink.net from my domain and I'm unable to also.

[someone_4]

in my sendmail.cf file the

# load average at which we just queue messages
#O QueueLA=8

# load average at which we refuse connections
#O RefuseLA=12

are at default..
Should I uncomment them double the value?

[Christopher Caldwell]

You should only adjust *LA parameters if load is an issue - i.e. if you run top, and the box is falling over, use *LA to control the load. By default, RefuseLA is 12, QueueLA is 8.

Make sure ident is set to 0:
#O Timeout.ident=30s
O Timeout.ident=0s