[SOLVED] DNS Caching / Lookup Problem?

[diggitydanc]

I recently started having messages that are being deferred because the remote connection is timing out. In each case it appears that the reason the connection is failing is related to the IP Address Zimbra is using for the remote system. An example:

zimbra@pp-colo:~/conf> grep F17D6ACCEB9 /var/log/zimbra.log
Oct 23 11:36:10 pp-colo postfix/smtpd[20796]: F17D6ACCEB9: client=localhost[127.0.0.1]
Oct 23 11:36:10 pp-colo postfix/cleanup[23659]: F17D6ACCEB9: message-id=<KLEIJIBDJODELLLDMNEBKEMACJAA.jnelson@pressaprint.com>
Oct 23 11:36:11 pp-colo postfix/qmgr[20072]: F17D6ACCEB9: from=<jnelson@pressaprint.com>, size=3563, nrcpt=1 (queue active)
Oct 23 11:36:11 pp-colo amavis[23566]: (23566-04) FWD via SMTP: <jnelson@pressaprint.com> -> <irelock@yahoo.com>,BODY=7BIT 250 2.6.0 Ok, id=23566-04, from MTA([127.0.0.1]:10025): 250 2.0.0 Ok: queued as F17D6ACCEB9
Oct 23 11:36:11 pp-colo amavis[23566]: (23566-04) Passed CLEAN, [72.164.193.147] <jnelson@pressaprint.com> -> <irelock@yahoo.com>, Message-ID: <KLEIJIBDJODELLLDMNEBKEMACJAA.jnelson@pressaprint.com>, mail_id: QLmhCuL6z4Rz, Hits: -2.587, size: 2859, queued_as: F17D6ACCEB9, 545 ms
Oct 23 11:36:11 pp-colo postfix/smtp[24865]: 6FB52ACCFA4: to=<irelock@yahoo.com>, relay=127.0.0.1[127.0.0.1]:10024, delay=0.8, delays=0.25/0/0/0.54, dsn=2.0.0, status=sent (250 2.0.0 Ok: queued as F17D6ACCEB9)
Oct 23 11:36:11 pp-colo postfix/qmgr[20072]: F17D6ACCEB9: to=<irelock@yahoo.com>, relay=none, delay=0.21, delays=0.11/0.11/0/0, dsn=4.4.1, status=deferred (delivery temporarily suspended: connect to yahoo.com[68.180.206.184]: Connection timed out)
Oct 23 11:43:01 pp-colo postfix/qmgr[20072]: F17D6ACCEB9: from=<jnelson@pressaprint.com>, size=3563, nrcpt=1 (queue active)
Oct 23 11:44:01 pp-colo postfix/smtp[4474]: F17D6ACCEB9: to=<irelock@yahoo.com>, relay=none, delay=470, delays=410/0.04/60/0, dsn=4.4.1, status=deferred (connect to yahoo.com[206.190.60.37]: Connection timed out)
Oct 23 11:53:00 pp-colo postfix/qmgr[20072]: F17D6ACCEB9: from=<jnelson@pressaprint.com>, size=3563, nrcpt=1 (queue active)
Oct 23 11:54:00 pp-colo postfix/smtp[13979]: F17D6ACCEB9: to=<irelock@yahoo.com>, relay=none, delay=1069, delays=1009/0.07/60/0, dsn=4.4.1, status=deferred (connect to yahoo.com[206.190.60.37]: Connection timed out)
Oct 23 12:13:02 pp-colo postfix/qmgr[20072]: F17D6ACCEB9: from=<jnelson@pressaprint.com>, size=3563, nrcpt=1 (queue active)
Oct 23 12:14:04 pp-colo postfix/smtp[12229]: F17D6ACCEB9: to=<irelock@yahoo.com>, relay=none, delay=2273, delays=2211/0.27/61/0, dsn=4.4.1, status=deferred (connect to yahoo.com[206.190.60.37]: Connection timed out)

Zimbra is attempting to connect to yahoo at 206.190.60.37. Proceed to dig mx records for yahoo.com:

zimbra@pp-colo:~/conf> dig mx yahoo.com

; <<>> DiG 9.4.1-P1 <<>> mx yahoo.com
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 48136
;; flags: qr rd ra; QUERY: 1, ANSWER: 7, AUTHORITY: 7, ADDITIONAL: 14

;; QUESTION SECTION:
;yahoo.com.                     IN      MX

;; ANSWER SECTION:
yahoo.com.              146     IN      MX      1 g.mx.mail.yahoo.com.
yahoo.com.              146     IN      MX      1 a.mx.mail.yahoo.com.
yahoo.com.              146     IN      MX      1 b.mx.mail.yahoo.com.
yahoo.com.              146     IN      MX      1 c.mx.mail.yahoo.com.
yahoo.com.              146     IN      MX      1 d.mx.mail.yahoo.com.
yahoo.com.              146     IN      MX      1 e.mx.mail.yahoo.com.
yahoo.com.              146     IN      MX      1 f.mx.mail.yahoo.com.

;; AUTHORITY SECTION:
yahoo.com.              165746  IN      NS      ns4.yahoo.com.
yahoo.com.              165746  IN      NS      ns5.yahoo.com.
yahoo.com.              165746  IN      NS      ns6.yahoo.com.
yahoo.com.              165746  IN      NS      ns8.yahoo.com.
yahoo.com.              165746  IN      NS      ns1.yahoo.com.
yahoo.com.              165746  IN      NS      ns2.yahoo.com.
yahoo.com.              165746  IN      NS      ns3.yahoo.com.

;; ADDITIONAL SECTION:
a.mx.mail.yahoo.com.    172     IN      A       209.191.118.103
b.mx.mail.yahoo.com.    172     IN      A       66.196.97.250
c.mx.mail.yahoo.com.    172     IN      A       216.39.53.3
c.mx.mail.yahoo.com.    172     IN      A       216.39.53.2
d.mx.mail.yahoo.com.    172     IN      A       66.196.82.7
e.mx.mail.yahoo.com.    172     IN      A       216.39.53.1
f.mx.mail.yahoo.com.    172     IN      A       209.191.88.247
f.mx.mail.yahoo.com.    172     IN      A       68.142.202.247
g.mx.mail.yahoo.com.    172     IN      A       209.191.88.239
g.mx.mail.yahoo.com.    172     IN      A       206.190.53.191
ns1.yahoo.com.          167821  IN      A       66.218.71.63
ns2.yahoo.com.          167821  IN      A       68.142.255.16
ns3.yahoo.com.          167821  IN      A       217.12.4.104
ns4.yahoo.com.          167821  IN      A       68.142.196.63

;; Query time: 2 msec
;; SERVER: 69.20.128.5#53(69.20.128.5)
;; WHEN: Thu Oct 23 12:48:02 2008
;; MSG SIZE  rcvd: 511

zimbra@pp-colo:~/conf>

The IP Address that Zimbra's attempting to connect to is not in the list of MX records for yahoo.com. The same is true of various messages to gmail, msn, aol, comcast, sbcglobal, etc (not isolated to a single provider). Here's my /etc/resolv.conf:

zimbra@pp-colo:~/conf> cat /etc/resolv.conf
nameserver 69.20.128.5
nameserver 69.20.129.5
search pressaprint.com

It's extremely basic and I think everything is right. This seems to have started in the last couple of days and is causing some frustration for my users.

For troubleshooting I've tried restarting zimbra and also flushing the postfix queue / reloading postifx, but so far to no effect. It's especially tough where it is random. Not all messages are affected, just some of them, but in each case if the message is in the deferred queue then it'll list an IP Address that's not currently in the MX Rotation as returned by dig.

Any advice or pointers would be invaluable and thanks in advance,

--Dan

[bdial]

that IP is actually the A record for yahoo.com, which is odd because mail servers usually don't try to use the A record unless it can't find any mx records but the nameservers you're using seem to have no problem. Is this happening on multiple domains? Do you have "enable dns lookups" checked on or off under the mta tab?

[diggitydanc]

I only have 1 domain set up in Zimbra. Recipients are spread across several domains including yahoo, msn, gmail, etc. I currently have 'enable dns lookups' off but will turn it on and requeue some of the messages as a test. I'll post back with the results here in a few minutes. Thanks for the reply,

--Dan

[diggitydanc]

Turning on 'Enable DNS Lookups' appears to have resolved the issue. My deferred queue cleared itself out nicely after I started requeuing messages in blocks.

Out of curiosity, why did / does Zimbra default to an A record instead of an MX record if that box is unchecked?

Thanks for your help,

--Dan