I have a new Mailman implementation connected to Postfix which is in turn connected to Dovecot. These two components are working correctly, as I have mailboxes set up with them and have created a test Mailman mailing list and received confirmation that it has in fact been created. Now I'm attempting to subscribe a test user in the normal way, by sending an email to {list-name}-subscribe@MyDomain.com. I have not received a subscription confirmation, but the following information from /var/log/mailman/error shows a problem:
Jul 02 14:15:01 2015 gate_news(18693): Traceback (most recent call last): Jul 02 14:15:01 2015 gate_news(18693): File "/usr/lib/mailman/cron/gate_news", line 293, in <module> Jul 02 14:15:01 2015 gate_news(18693): main() Jul 02 14:15:01 2015 gate_news(18693): File "/usr/lib/mailman/cron/gate_news", line 268, in main Jul 02 14:15:01 2015 gate_news(18693): lock.lock(timeout=0.5) Jul 02 14:15:01 2015 gate_news(18693): File "/usr/lib/mailman/Mailman/LockFile.py", line 243, in lock Jul 02 14:15:01 2015 gate_news(18693): self.__write() Jul 02 14:15:01 2015 gate_news(18693): File "/usr/lib/mailman/Mailman/LockFile.py", line 422, in __write Jul 02 14:15:01 2015 gate_news(18693): fp = open(self.__tmpfname, 'w') Jul 02 14:15:01 2015 gate_news(18693): IOError : [Errno 13] Permission denied: '/var/lock/mailman/gate_news.lock.{my-system-name}.{MyDomain.com}.18693.0' Here are the directory permissions:
drwxrwsr-x 2 root mailman 80 Jul 2 13:57 mailman lrwxrwxrwx 1 root root 11 Sep 14 2012 lock -> ../run/lock
I also receive messages like these about every fifteen minutes in /var/log/mailman/bounce:
Jul 02 14:42:12 2015 (1308) post to mailman from mailman-owner@{my-domain.com}, size=1736, message-id=<mailman.53.1435726803.13146.mailman@{my-domain.com}>, 1 failures Jul 02 14:42:16 2015 (1308) post to mailman from mailman-owner@{my-domain.com}, size=1733, message-id=<mailman.98.1435726804.13146.mailman@{my-domain.com}>, 1 failures
Thanks in advance for any assistance. I suspect it's something I've overlooked.
On 07/02/2015 07:56 AM, Steve Matzura wrote:
gate_news runs every 5 minutes. do you see this error every 5 minutes? Have you run Mailman's bin/check_perms?
In any case, issues with gate_news only affect gating from usenet of those lists that have Mail<->News gateways -> gateway_to_mail = Yes.
Mailman is sending a notice to the 'mailman' list and it's undeliverable. That doesn't look like a 'bounce' log message. It looks more like 'post' log. What's in the 'smtp-failure' log?
Thanks in advance for any assistance. I suspect it's something I've overlooked.
Yes. What's in the Postfix logs for these failures and for the message sent to {list-name}-subscribe@MyDomain.com?
The most common issue here is you are using the Dovecot LDA and it doesn't know how to deliver to Mailman. If you are using aliases to deliver to Mailman, you have to ensure that mail to Mailman is handles by the Postfix 'local' LDA, not Dovecot. See the FAQ at <http://wiki.list.org/x/10715238> for a way to do this.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Thu, 02 Jul 2015 08:23:23 -0700, you wrote:
I just ran check_perms to fix another problem, and it appears the two *were* related, as this one's now gone as well.
In any case, issues with gate_news only affect gating from usenet of those lists that have Mail<->News gateways -> gateway_to_mail = Yes.
Mine don't, or shouldn't. If they do, I'll turn that off immediately.
The smpt log is now clean. Newest message is ten minutes older than the current date and time. I suspect running check_perms fixed everything.
On 07/02/2015 09:07 AM, Steve Matzura wrote:
Good. Incorrect permissions can cause many issues.
I'm sure they don't, but cron/gate_news still runs and checks to see if any lists have newsgroups that might need polling. If you won't ever do any usenet gatewaying, you can comment out or remove the gate_news entry in Mailman's crontab.
Does a 'subscribe' email get delivered to Mailman? That could have been a permissions issue, but it's usually something else.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Thu, 02 Jul 2015 08:23:23 -0700, you wrote:
Jul 02 16:12:13 2015 (1308) <mailman.53.1435726803.13146.mailman@{my-domain.com}> smtp to mailman for 1 recips, completed in 4.225 seconds Jul 02 16:12:17 2015 (1308) <mailman.98.1435726804.13146.mailman@{my-domain.com}> smtp to mailman for 1 recips, completed in 4.159 seconds
I get this about every fifteen minutes.
There is one additional wrinkle here. I have two systems, one production, one test. They are near duplicates of each other except for age of software--production's old, test is near-current. Haven't installed MM 2.1.20 yet. It would be all right if the messages to mailman@{my-domain.com} went to the wrong machine, but they don't seem to be going anywhere.
On 07/02/2015 09:23 AM, Steve Matzura wrote:
And those look like 'smtp' log messages, not 'smtp-failure'
And once again, what's in the Postfix logs for these failures and for the message sent to {list-name}-subscribe@MyDomain.com?
If messages to Mailman are being routed to the Dovecot transport, they won't be properly delivered. If you are using aliases, Mailman's mail has to be routed to the 'local' transport.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Thu, 02 Jul 2015 16:46:00 -0700, you wrote:
This sequence occurs every fifteen minutes:
Jul 03 09:57:21 2015 (1308) All recipients refused: {'kelly@{mydomain.com}': (450, '4.1.1 <kelly@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table')}, msgid: <mailman.53.1435726803.13146.mailman@{mydomain.com}> Jul 03 09:57:21 2015 (1308) delivery to kelly@{mydomain.com} failed with code 450: 4.1.1 <kelly@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table Jul 03 09:57:25 2015 (1308) All recipients refused: {'lynn@{mydomain.com}': (450, '4.1.1 <lynn@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table')}, msgid: <mailman.98.1435726804.13146.mailman@{mydomain.com}> Jul 03 09:57:25 2015 (1308) delivery to lynn@{mydomain.com} failed with code 450: 4.1.1 <lynn@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table
On 07/03/2015 03:17 AM, Steve Matzura wrote:
Mailman is trying to send some notices to 'kelly' and 'lynn' and Postfix is not accepting them with a 450 status. Since it is a 4xx retryable status, Mailman keeps retrying every 15 minutes. Mailman will eventually give up and return bounces after 5 days (default DELIVERY_RETRY_PERIOD).
In order to help you proceed, I'd like to see the output from 'postconf -n' and the postfix log message from one of the above rejects.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On 07/03/2015 07:01 AM, Mark Sapiro wrote:
Note that the above messages are not really Mailman related. They don't indicate that Postfix can't deliver to Mailman, although that may also be true. They indicate that Postfix cannot deliver to 'lynn' or 'kelly'. If these should be valid, deliverable addresses, there is a Postfix configuration issue. If they are in fact invalid, you can check your main.cf settings for things like 'soft_bounce' and 'unknown_virtual_mailbox_reject_code'. You may want 'soft_bounce = no' and unknown_virtual_mailbox_reject_code = 550.
Also, to get rid of these retries before they expire, you can remove the four $prefix/qfiles/retry/*.pck files. You can also do
$prefix/bin/show_qfiles $prefix/qfiles/retry/*.pck
to see what the messages are.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On 07/02/2015 07:56 AM, Steve Matzura wrote:
gate_news runs every 5 minutes. do you see this error every 5 minutes? Have you run Mailman's bin/check_perms?
In any case, issues with gate_news only affect gating from usenet of those lists that have Mail<->News gateways -> gateway_to_mail = Yes.
Mailman is sending a notice to the 'mailman' list and it's undeliverable. That doesn't look like a 'bounce' log message. It looks more like 'post' log. What's in the 'smtp-failure' log?
Thanks in advance for any assistance. I suspect it's something I've overlooked.
Yes. What's in the Postfix logs for these failures and for the message sent to {list-name}-subscribe@MyDomain.com?
The most common issue here is you are using the Dovecot LDA and it doesn't know how to deliver to Mailman. If you are using aliases to deliver to Mailman, you have to ensure that mail to Mailman is handles by the Postfix 'local' LDA, not Dovecot. See the FAQ at <http://wiki.list.org/x/10715238> for a way to do this.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Thu, 02 Jul 2015 08:23:23 -0700, you wrote:
I just ran check_perms to fix another problem, and it appears the two *were* related, as this one's now gone as well.
In any case, issues with gate_news only affect gating from usenet of those lists that have Mail<->News gateways -> gateway_to_mail = Yes.
Mine don't, or shouldn't. If they do, I'll turn that off immediately.
The smpt log is now clean. Newest message is ten minutes older than the current date and time. I suspect running check_perms fixed everything.
On 07/02/2015 09:07 AM, Steve Matzura wrote:
Good. Incorrect permissions can cause many issues.
I'm sure they don't, but cron/gate_news still runs and checks to see if any lists have newsgroups that might need polling. If you won't ever do any usenet gatewaying, you can comment out or remove the gate_news entry in Mailman's crontab.
Does a 'subscribe' email get delivered to Mailman? That could have been a permissions issue, but it's usually something else.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Thu, 02 Jul 2015 08:23:23 -0700, you wrote:
Jul 02 16:12:13 2015 (1308) <mailman.53.1435726803.13146.mailman@{my-domain.com}> smtp to mailman for 1 recips, completed in 4.225 seconds Jul 02 16:12:17 2015 (1308) <mailman.98.1435726804.13146.mailman@{my-domain.com}> smtp to mailman for 1 recips, completed in 4.159 seconds
I get this about every fifteen minutes.
There is one additional wrinkle here. I have two systems, one production, one test. They are near duplicates of each other except for age of software--production's old, test is near-current. Haven't installed MM 2.1.20 yet. It would be all right if the messages to mailman@{my-domain.com} went to the wrong machine, but they don't seem to be going anywhere.
On 07/02/2015 09:23 AM, Steve Matzura wrote:
And those look like 'smtp' log messages, not 'smtp-failure'
And once again, what's in the Postfix logs for these failures and for the message sent to {list-name}-subscribe@MyDomain.com?
If messages to Mailman are being routed to the Dovecot transport, they won't be properly delivered. If you are using aliases, Mailman's mail has to be routed to the 'local' transport.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Thu, 02 Jul 2015 16:46:00 -0700, you wrote:
This sequence occurs every fifteen minutes:
Jul 03 09:57:21 2015 (1308) All recipients refused: {'kelly@{mydomain.com}': (450, '4.1.1 <kelly@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table')}, msgid: <mailman.53.1435726803.13146.mailman@{mydomain.com}> Jul 03 09:57:21 2015 (1308) delivery to kelly@{mydomain.com} failed with code 450: 4.1.1 <kelly@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table Jul 03 09:57:25 2015 (1308) All recipients refused: {'lynn@{mydomain.com}': (450, '4.1.1 <lynn@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table')}, msgid: <mailman.98.1435726804.13146.mailman@{mydomain.com}> Jul 03 09:57:25 2015 (1308) delivery to lynn@{mydomain.com} failed with code 450: 4.1.1 <lynn@{mydomain.com}>: Recipient address rejected: User unknown in virtual mailbox table
On 07/03/2015 03:17 AM, Steve Matzura wrote:
Mailman is trying to send some notices to 'kelly' and 'lynn' and Postfix is not accepting them with a 450 status. Since it is a 4xx retryable status, Mailman keeps retrying every 15 minutes. Mailman will eventually give up and return bounces after 5 days (default DELIVERY_RETRY_PERIOD).
In order to help you proceed, I'd like to see the output from 'postconf -n' and the postfix log message from one of the above rejects.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On 07/03/2015 07:01 AM, Mark Sapiro wrote:
Note that the above messages are not really Mailman related. They don't indicate that Postfix can't deliver to Mailman, although that may also be true. They indicate that Postfix cannot deliver to 'lynn' or 'kelly'. If these should be valid, deliverable addresses, there is a Postfix configuration issue. If they are in fact invalid, you can check your main.cf settings for things like 'soft_bounce' and 'unknown_virtual_mailbox_reject_code'. You may want 'soft_bounce = no' and unknown_virtual_mailbox_reject_code = 550.
Also, to get rid of these retries before they expire, you can remove the four $prefix/qfiles/retry/*.pck files. You can also do
$prefix/bin/show_qfiles $prefix/qfiles/retry/*.pck
to see what the messages are.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
participants (3)
-
Mark Sapiro -
Steve Matzura -
Steve Matzura