Mailing List Archive: DBMail: users

dbmail-replycache table / vacation loop

Index | Next | Previous | View Threaded

Jon at host-it

Jun 25, 2009, 12:20 PM

Post #1 of 6 (386 views)
Permalink

dbmail-replycache table / vacation loop

Hi Guys,

We have what looks like looping mail between two auto responders (by way of vacation in sieve).

I can see in our lmtpd logs lots of failed inserts in teh dbmail_replycache table, such as;

Jun 25 20:07:41 mail4-db-2 dbmail/lmtpd[427]: Error:[sql] dbmysql.c,db_query(+290): [Duplicate entry 'Customer.Care[at]vodafone.co.uk-donotreply@somedomain.c' for key 1] [INSERT INTO dbmail_replycache (to_addr, from_addr, handle, lastseen) VALUES ('

Having emptied the users mailbox we're seeing the same email hit the mailbox and have since resorted to rejecting @ postfix.

Any ideas on this one? I'm sure this is a recent issue and nothing notable has changed since we went up to 2.2.11 some months ago.

Is this a bug? We're running 2.2.11 on both lmtpd and sieve.

We run dbmail-util -r 72h twice daily also.

Thanks

Jon

paul at nfg

Jun 25, 2009, 2:08 PM

Post #2 of 6 (356 views)
Permalink

Re: dbmail-replycache table / vacation loop [In reply to]

Jon Duggan wrote:
> Jun 25 20:07:41 mail4-db-2 dbmail/lmtpd[427]: Error:[sql]
> dbmysql.c,db_query(+290): [Duplicate entry
> 'Customer.Care[at]vodafone.co.uk-donotreply@somedomain.c' for key 1]
> [INSERT INTO dbmail_replycache (to_addr, from_addr, handle, lastseen)
> VALUES ('

Jon,

I would need to see the full query, and the one just before this one.
You'll need to run at trace_errlog=5 to get at them, or use your mysql
logs if you have them.

> Having emptied the users mailbox we’re seeing the same email hit the
> mailbox and have since resorted to rejecting @ postfix.

This might happen if the same message is being inserted twice,
simulaneously, be two different lmtp processes.

> Is this a bug? We’re running 2.2.11 on both lmtpd and sieve.

How often do you see this?

>
>
>
> We run dbmail-util –r 72h twice daily also.

That's silly. Way too short, way too often. Once a day, with -r 240
should be better. If you set -r too small, the days value in the sieve
scripts is effectively being capped.

>
>
>
> Thanks
>
>
>
> Jon
>
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> DBmail mailing list
> DBmail[at]dbmail.org
> http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail

--
________________________________________________________________
Paul Stevens paul at nfg.nl
NET FACILITIES GROUP GPG/PGP: 1024D/11F8CD31
The Netherlands________________________________http://www.nfg.nl
_______________________________________________
DBmail mailing list
DBmail[at]dbmail.org
http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail

Jon at host-it

Jun 25, 2009, 2:21 PM

Post #3 of 6 (355 views)
Permalink

Re: dbmail-replycache table / vacation loop [In reply to]

Thanks paul,

> I would need to see the full query, and the one just before this one.
> You'll need to run at trace_errlog=5 to get at them, or use your mysql
> logs if you have them.

I'll get some better logs and post to the list.

> This might happen if the same message is being inserted twice,
> simulaneously, be two different lmtp processes.

>> Is this a bug? We’re running 2.2.11 on both lmtpd and sieve.

> How often do you see this?

I'm seeing this on a daily basis, for the last few days. An example today is two autoresponders bounced over 5,000 emails to eachother through the course of today, until i found the logs and stopped it. (where the autoresponder was to some @vodafone.co.uk address, so they obviously also have problems)

>>
>>
>>
>> We run dbmail-util –r 72h twice daily also.

> That's silly. Way too short, way too often. Once a day, with -r 240
> should be better. If you set -r too small, the days value in the sieve
> scripts is effectively being capped.

As i understood the usage, it will remove replycache info older than 72hours. This to me seems a sane figure for old entries to be removed? Am i misunderstanding its use? I've only recently started using -r 72h having found that the table was about 1/2 million rows and realised it was not being cleaned.

Jon

_______________________________________________
DBmail mailing list
DBmail[at]dbmail.org
http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail

aaron at serendipity

Jun 25, 2009, 6:20 PM

Post #4 of 6 (359 views)
Permalink

Re: dbmail-replycache table / vacation loop [In reply to]

On Thu, 25 Jun 2009 22:21:30 +0100, Jon Duggan <Jon[at]host-it.co.uk> wrote:
> Thanks paul,
>
>
>> I would need to see the full query, and the one just before this one.
>> You'll need to run at trace_errlog=5 to get at them, or use your mysql
>> logs if you have them.
>
> I'll get some better logs and post to the list.
>
>> This might happen if the same message is being inserted twice,
>> simulaneously, be two different lmtp processes.
>
>>> Is this a bug? We’re running 2.2.11 on both lmtpd and sieve.
>
>> How often do you see this?
>
> I'm seeing this on a daily basis, for the last few days. An example
today
> is two autoresponders bounced over 5,000 emails to eachother through the
> course of today, until i found the logs and stopped it. (where the
> autoresponder was to some @vodafone.co.uk address, so they obviously also
> have problems)

What version of libSieve do you have? There have been some additional
checks for autoresponder headers that might be relevant here.

Aaron
_______________________________________________
DBmail mailing list
DBmail[at]dbmail.org
http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail

paul at nfg

Jun 26, 2009, 12:36 AM

Post #5 of 6 (352 views)
Permalink

Re: dbmail-replycache table / vacation loop [In reply to]

Jon Duggan wrote:

> As i understood the usage, it will remove replycache info older than
> 72hours. This to me seems a sane figure for old entries to be
> removed? Am i misunderstanding its use? I've only recently started
> using -r 72h having found that the table was about 1/2 million rows
> and realised it was not being cleaned.

Nope, you misunderstood. Indeed the replycache will be cleaned of
entries where lastseen is before 72hours before now. Which you don't
want, really. If you use something like days: 7 in your scripts, you
don't want to send a reply message from the same sieve script to the
same sender more than once every seven days. But with a -r 72h, you will
truncate the replycache forcing a days: 3

--
________________________________________________________________
Paul Stevens paul at nfg.nl
NET FACILITIES GROUP GPG/PGP: 1024D/11F8CD31
The Netherlands________________________________http://www.nfg.nl
_______________________________________________
DBmail mailing list
DBmail[at]dbmail.org
http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail

paul at nfg

Jun 26, 2009, 12:38 AM

Post #6 of 6 (353 views)
Permalink

Re: dbmail-replycache table / vacation loop [In reply to]

Aaron Stone wrote:

> What version of libSieve do you have? There have been some additional
> checks for autoresponder headers that might be relevant here.

Looking at the level=5 logs Jon sent to me offlist I see the whole reply
message being used as handle. So either Jon is using the reply message
as a handle: in his script, or dm_md5 is broken.

--
________________________________________________________________
Paul Stevens paul at nfg.nl
NET FACILITIES GROUP GPG/PGP: 1024D/11F8CD31
The Netherlands________________________________http://www.nfg.nl
_______________________________________________
DBmail mailing list
DBmail[at]dbmail.org
http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact lists@gossamer-threads.com