deliver reports (/var/lib/imap/socket/lmtp) failed: Connection refused

Gyorgy Knyihar cyrus123 at syspro.hu
Sat Sep 22 02:08:16 EDT 2007


Hello Simon,

I reconstructed all mailboxes but it did not help.
I found that the service outages happen usually during nights.
There were two backups run by cron every night which gzipped a lot of data.
I disabled these CPU intensive backups and the cyrus processes are  
stable for more than 6 days now. I can understand why pop3s, imaps  
request timed out during the CPU sensitive backups but I still don't  
understand why only restart helped. I tried to run the backups with  
lower priority using nice but it did not help.

Thanks for your help.

Regards, Gyorgy

Idézet (Gyorgy Knyihar <cyrus123 at syspro.hu>):

> Hello Simon,
>
> Thank you very much for your help.
> Please see my answers below.
>
> Idezet (Simon Matter <simon.matter at invoca.ch>):
>
>>> Hello Simon,
>>>
>>> I tried you suggestion and there are no more
>>> "Sep  3 07:45:59 srv1 deliver[20650]:
>>> connect(/var/lib/imap/socket/lmtp) failed: Connection refused"
>>> messages appearing in the error log file.
>>>
>>> Many thanks for your help.
>>>
>>> I run postfix chrooted so a hardlink pointing to
>>> /var/lib/imap/socket/lmtp was required in
>>> /var/spool/postfix/var/lib/imap/socket directory. This needs to be
>>> refreshed every time cyrus is restarted.
>>
>> It's much easier to make lmtp not chrooted. Simply remove the chroot
>> option from lmtp service in master.cf. I don't think that's a big security
>> problem.
>
> Yes, this sounds easier, thanks.
>
>>
>>>
>>> But my collegaues told me that this is not the only problem. Cyrus
>>> pop3, pop3s, imap and imaps services become unavailable time to time.
>>> And only restart helps.
>>> Delivery from postfix to cyrus mailboxes via this new setup (direct
>>> lmtp) is OK during the outage of pop3, pop3s, imap, imaps service. The
>>> services don't die at the same time. Sometimes imap dies but pop3
>>> still works. Restart helps. It is really strange. It was a stable
>>> system before and nothing was changed. I could find only the following
>>> errormessages in the log files:
>>>
>>> Sep  8 01:35:00 srv1 pop3s[15591]: Fatal error: tls_start_servertls()
>>> failed
>>> Sep  8 01:37:09 srv1 imaps[15687]: Fatal error: tls_start_servertls()
>>> failed
>>
>> Do you only have problems with pop3s and imaps, or also with pop3 and
>> imap? If it's only SSL/TLS, there yould be a problem with /dev/random and
>> you should try using /dev/urandom instead. Otherwise it could be that one
>> or more of your cyrus databases are corrupt. Did you try a reconstruct on
>> all you mailboxes?
>
> It is a problem with pop3 and imap as well. I monitor the services
> with nagios running on the same machine and there are problems with
> all services.
> Time to time these services die and only restart helps.
> I haven't tried to reconstruct mailboxes yet. I will try.
> I don't know if it does matter but lot of users are over their quota.
>
> Thanks a lot for your help.
>
> Regards, Gyorgy
>
>>
>> Simon
>>
>>>
>>> and
>>>
>>> Sep  8 02:17:47 srv1 cyrus-master[20423]: pop3 has -5 workers?!?
>>>
>>> There are many of such messages. I can access my mailbox using TLS
>>> without any problem.
>>>
>>> Any help would be appreciated.
>>>
>>> Thanks, Gyorgy
>>>
>>> Idézet (Simon Matter <simon.matter at invoca.ch>):
>>>
>>>>> Hello,
>>>>>
>>>>> We had a properly working postfix + cyrus-imap system till 23rd of
>>>>> July. Then messages like the below one started to appear in the log
>>>>> file.
>>>>>
>>>>> Sep  3 07:45:59 srv1 deliver[20650]:
>>>>> connect(/var/lib/imap/socket/lmtp) failed: Connection refused
>>>>>
>>>>> The postfix log shows:
>>>>>
>>>>> Sep  3 07:45:59 srv1 postfix/pipe[19972]: B88D01DC8EF8: to=<xxx@
>>>>> srv1.xxx.xx>, orig_to=<xxx at xxx.xx>, relay=cyrus-deliver, delay=2
>>>>> 49424, status=deferred (temporary failure. Command output: couldn't
>>>>> connect to l
>>>>> mtpd: Connection refused_ 421 4.3.0 deliver: couldn't connect to lmtpd_
>>>>> )
>>>>>
>>>>>
>>>>> Do you have any ideas what could happen? It worked fine before.
>>>>>
>>>>>
>>>>> cyrus.conf contains:
>>>>>
>>>>>    # at least one LMTP is required for delivery
>>>>> #  lmtp         cmd="lmtpd" listen="lmtp" prefork=0
>>>>>    lmtpunix      cmd="lmtpd" listen="/var/lib/imap/socket/lmtp"
>>>>> prefork=5
>>>>>
>>>>>
>>>>> postfix/master.cf contains:
>>>>>
>>>>> cyrus-deliver     unix  -       n       n       -       -       pipe
>>>>>    user=cyrus argv=/usr/lib/cyrus-imapd/deliver -e -r ${sender} -m
>>>>> ${extension} ${user}
>>>>>
>>>>>
>>>>> postfix/main.cf contains:
>>>>>
>>>>> mailbox_transport = cyrus-deliver
>>>>
>>>> Hi,
>>>>
>>>> I don't know why things suddenly fail but you shouldn't use the cyrus
>>>> transport anymore. You should use lmtp as a mailtransport directly. If
>>>> your postfix setup isn't somehow exotic you should be able to simply
>>>> switch with this config:
>>>> mailbox_transport = lmtp:unix:/var/lib/imap/socket/lmtp
>>>>
>>>> Maybe you could give it a try.
>>>>
>>>> Simon
>>>>
>>>>
>>>
>>>
>>>
>>> ----
>>> Cyrus Home Page: http://cyrusimap.web.cmu.edu/
>>> Cyrus Wiki/FAQ: http://cyrusimap.web.cmu.edu/twiki
>>> List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html
>>
>
>
>
> ----
> Cyrus Home Page: http://cyrusimap.web.cmu.edu/
> Cyrus Wiki/FAQ: http://cyrusimap.web.cmu.edu/twiki
> List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html





More information about the Info-cyrus mailing list