Error Sending

[dmoring]

Installed on FC 4 and running/receiving messages fine. However, when a message is attempted to be sent an error is thrown (below). I know it is not the connection as I can telnet to that domains mx from the Zimbra machine on port 25 and talk fine with the other mail server.

msg - system failure: MessagingException
code - service.FAILURE
method - ZmCsfeCommand.invoke
detail - soap:Receiver
constructor -
function ZmCsfeException(msg, code, method, detail) {
if (arguments.length == 0) {
return;
}
AjxException.call(this, msg, code, method, detail);
}

toString -
function () {
return "ZmCsfeException";
}

dump -
function () {
return "AjxException: msg=" + this.msg + " code=" + this.code + " method=" + this.method + " detail=" + this.detail;
}

[anand]

The full stack trace/exception message in /opt/zimbra/log/zimbra.log should have more detail on why SMTP send from the tomcat process -> postfix is failing.

[dmoring]

Ah... OK.... Here appears to be the germain part of the entry:
SendMsg - exception ocurred during SendMsg
com.sun.mail.smtp.SMTPSendFailedException: 451 Error: queue file write error

[dmoring]

Please ignore previous post... wrong error. Error is posted below. Here's the funny thing... the message gets through! After completion, the remote server issues the 250 and 221 Quit ack, but Zimbra appears to timeout waiting...

Caused by: javax.mail.MessagingException: Exception reading response;
nested exception is:
java.net.SocketTimeoutException: Read timed out
at com.sun.mail.smtp.SMTPTransport.readServerResponse (SMTPTransport.java:1349)
at com.sun.mail.smtp.SMTPTransport.issueSendCommand(S MTPTransport.java:1263)
at com.sun.mail.smtp.SMTPTransport.finishData(SMTPTra nsport.java:1149)
at com.sun.mail.smtp.SMTPTransport.sendMessage(SMTPTr ansport.java:527)
at javax.mail.Transport.send0(Transport.java:151)
at javax.mail.Transport.send(Transport.java:80)
at com.zimbra.cs.service.mail.SendMsg.sendMimeMessage (SendMsg.java:253)
... 25 more

[anand]

which SMTP server is tomcat trying to connect to? What is your setting
of zimbraSmtpHostname and zimbraSmtpPort?

$ zmprov GetAllServers
<your-server-name>
$ zmprov GetServer <your-server-name> | grep Smtp

Looks like tomcat+javamail is waiting for a 250 after DATA command
and is not getting it? Are you making your "is sending 250/221" conclusions
from tcpdump?

[dmoring]

which SMTP server is tomcat trying to connect to? What is your setting
of zimbraSmtpHostname and zimbraSmtpPort?

$ zmprov GetAllServers
<your-server-name>
$ zmprov GetServer <your-server-name> | grep Smtp

The setting is the local host name (linhost.tmcentral.net) and the Zimbra smtp port is port 25. It appears to occur on any given remote smtp server (tested using exchange and mdaemon as the remote servers).

Looks like tomcat+javamail is waiting for a 250 after DATA command
and is not getting it? Are you making your "is sending 250/221" conclusions
from tcpdump?

No, I am making them from a remote server's log (to which I have access) of the smtp session between it and the Zimbra host box. I can also confirm that all tested remote servers accept the mail and send the 250/221 commands (and deliver the mail to their respective boxes). Any other ideas?

[anand]

The message path in your setup should be:

tomcat (zimbra/localhost) -> postfix (zimbra/localhost) -> exchange (remote host)

We should now check that the middle piece is happy.

Check /var/log/maillog (FYI: in the future we will have zimbra installed postfix log to /var/log/zimbra.log - log consolidation).

Looks to me like the postfix is unhappy for some reason and tomcat is timing out on postfix. There should be something useful in the postfix logs.

(PS: on a related note, did you set relay host in the zimbra admin console so postfix can relay to exchange? if you set in the admin console, we have a periodic script than scans LDAP and pushed postfix config, so after a certain delay you should see postfix pick up the change - the reason I bring this up is because postfix will wedge if it thinks mail loops to itself etc - because of a misconfiguration.)

[dmoring]

The message path in your setup should be:

tomcat (zimbra/localhost) -> postfix (zimbra/localhost) -> exchange (remote host)

We should now check that the middle piece is happy.

Check /var/log/maillog (FYI: in the future we will have zimbra installed postfix log to /var/log/zimbra.log - log consolidation).

Here's a funny thing... I just tried to send again and it worked flawlessly! Not really sure whats going on since I didn't change anything! Nevertheless, I'm going to compare the postfix log you note with the zimbra.log and will post the results where it errored, just to understand what went wrong!

(PS: on a related note, did you set relay host in the zimbra admin console so postfix can relay to exchange? if you set in the admin console, we have a periodic script than scans LDAP and pushed postfix config, so after a certain delay you should see postfix pick up the change - the reason I bring this up is because postfix will wedge if it thinks mail loops to itself etc - because of a misconfiguration.)

I did not set any relaying/"Smart Host" up at all.; should I have? This is a lab with another server on the local subnet that has no issue sending (all RFC reqs are met for the IP) so I didn't see any need to have a backup relay for testing (after all, in testing my goal is to make sure the native process works; the email contents are garbage at this point anyway!)

[marcmac]

I did not set any relaying/"Smart Host" up at all.; should I have? This is a lab with another server on the local subnet that has no issue sending (all RFC reqs are met for the IP) so I didn't see any need to have a backup relay for testing (after all, in testing my goal is to make sure the native process works; the email contents are garbage at this point anyway!)

You don't need a smart host, unless the DNS lookups from the ZCS server aren't working - if postfix can't find the next hop (which doesn't appear to have been the case) it won't send the message.

[kplau]

I had similar problems just as described and the logs were all similar, timeout and yeap it was the fact postfix wasn't happy.

In this case it was my fault - if you ever find yourself managing distribution lists and adding mailing list protections (which imvho should have been done via a gui) - and if you ever mess with the internal files as I did (there wasn't any other way anyhow) - as described here - RestrictPostfixRecipients - ZimbraWiki..

Don't..ever.. remove.. the .db and permitted senders list file on its own - you will have to retrace all your steps and remove all the referals to these files such as those in the script /opt/zimbra/postfix/conf/update_protected_recipients as well as in /opt/zimbra/postfix/conf/main.cf

Gave me quite a scare this morning as I know nothing about email servers nor how to troubleshoot them!

p/s:dont forget to reload postfix after that