Postfix active queue builds up and not get cleaned

Happy New to everybody, friends!

Hope somebody could shed me some light on my phenomenon.

I have two identical machines running Maia. They work in
parallel, i.e. two
MX records point to them as say mail02.domain.tld and
mail03.domain.tld.
Both of them are FreeBSD 6.2, same hardware.
Both run postfix 2.4-20061006.

There's nagging difference in the way one of the machines
handles mail,
namely mail in the queue.
I noticed that mail accumulates in /var/spool/postfix/active
folder. Indeed,
running mailq shows that mail is present in the queue and
not only does it
stay there not being sent but it grows. Sometimes it helps
if you run
"postfix flush" but in essence you are not
supposed to babysit postfix
looking at its queue and flushing it. After a day and night
the queue size
grows to Megabytes. I used to do a trick, moving active
folder with all
emails stuck in the queue to the other machine and this
machine successfully
sends everything away without any problems.

I started analyze the situation looking into logs but
nothing revealed me
any clues leading to this problem. Looks like the queue
manager is working,
I see this lines in the maillog file:

Dec 28 10:08:10 mail03 postfix/qmgr[17701]: 56CF71AF421:
removed

Is there any else that could help me find a solution to it?

Sincerely
Eugene



_______________________________________________
Maia-users mailing list
Maia-usersrenaissoft.com
http://www.renaissoft.com/mailman/listinfo/maia-users

Eugene Pefti wrote:
> Happy New to everybody, friends!
>
> Hope somebody could shed me some light on my
phenomenon.
>
> I have two identical machines running Maia. They work
in parallel, i.e. two
> MX records point to them as say mail02.domain.tld and
mail03.domain.tld.
> Both of them are FreeBSD 6.2, same hardware.
> Both run postfix 2.4-20061006.
>
> There's nagging difference in the way one of the
machines handles mail,
> namely mail in the queue.
> I noticed that mail accumulates in
/var/spool/postfix/active folder. Indeed,
> running mailq shows that mail is present in the queue
and not only does it
> stay there not being sent but it grows. Sometimes it
helps if you run
> "postfix flush" but in essence you are not
supposed to babysit postfix
> looking at its queue and flushing it. After a day and
night the queue size
> grows to Megabytes. I used to do a trick, moving active
folder with all
> emails stuck in the queue to the other machine and this
machine successfully
> sends everything away without any problems.
>
> I started analyze the situation looking into logs but
nothing revealed me
> any clues leading to this problem. Looks like the queue
manager is working,
> I see this lines in the maillog file:
>
> Dec 28 10:08:10 mail03 postfix/qmgr[17701]:
56CF71AF421: removed
>
> Is there any else that could help me find a solution to
it?
>   

Where is amavisd-maia running?  both machines? somewhere
else?

Do both of these machines then forward to somewhere else?

look for any messages about connections being refused... 
Make sure you
have enough amavisd-maia children for your postfix servers -
if they are
sharing, one could be starving the other.
_______________________________________________
Maia-users mailing list
Maia-usersrenaissoft.com
http://www.renaissoft.com/mailman/listinfo/maia-users

Hi David,

Thanks for an answer. The only thing that both machine share
is MySQL
database. It is hosted on the other box. I don't see any
problem connecting
to it. At least I see successful connections to db in
amavisd log file.
Amavisd-maia is running on every so-called frond-end
machine. The machine in
question has enough resources to handle active mail, 600MB
of RAM is always
free out of 2GB available.

Both machines have identical relay_domains and transport
files and after
receiving mail back from amavisd they forward mail to
downstream mail
servers

This is part of maillog file associated with handling of one
of emails
sitting in the queue, in /var/spool/postfix/active:
(024BD1AF459 is message
ID)

Dec 28 10:54:48 mail03 postfix/smtpd[18051]: connect from
idcmail-mo1so.shaw.ca[24.71.223.10]                         
      
Dec 28 10:54:49 mail03 postfix/smtpd[18051]: 024BD1AF459:
client=idcmail-mo1so.shaw.ca[24.71.223.10]                  
      
Dec 28 10:54:49 mail03 postfix/cleanup[18069]: 024BD1AF459:
message-id=<0JTR00FHTWFM6660l-daemon>               
           
Dec 28 10:54:49 mail03 postfix/qmgr[17701]: 024BD1AF459:
from=<elsamanshaw.ca>, size=3449, nrcpt=1 (queue
active)           
Dec 28 10:54:49 mail03 postfix/smtpd[18051]: disconnect
from
idcmail-mo1so.shaw.ca[24.71.223.10]

As we see, there's only one reference saying the it was put
to the active
queue and that's all, nothing about this message ID.  

Eugene

-----Original Message-----
From: David Morton [mailto:mortondadgrmm.net] 
Sent: Friday, December 28, 2007 10:38 AM
To: Eugene Pefti
Cc: maia-usersrenaissoft.com
Subject: Re: [Maia-users] Postfix active queue builds up and
not get cleaned

Eugene Pefti wrote:
> Happy New to everybody, friends!
>
> Hope somebody could shed me some light on my
phenomenon.
>
> I have two identical machines running Maia. They work
in parallel, i.e.
two
> MX records point to them as say mail02.domain.tld and
mail03.domain.tld.
> Both of them are FreeBSD 6.2, same hardware.
> Both run postfix 2.4-20061006.
>
> There's nagging difference in the way one of the
machines handles mail,
> namely mail in the queue.
> I noticed that mail accumulates in
/var/spool/postfix/active folder.
Indeed,
> running mailq shows that mail is present in the queue
and not only does it
> stay there not being sent but it grows. Sometimes it
helps if you run
> "postfix flush" but in essence you are not
supposed to babysit postfix
> looking at its queue and flushing it. After a day and
night the queue size
> grows to Megabytes. I used to do a trick, moving active
folder with all
> emails stuck in the queue to the other machine and this
machine
successfully
> sends everything away without any problems.
>
> I started analyze the situation looking into logs but
nothing revealed me
> any clues leading to this problem. Looks like the queue
manager is
working,
> I see this lines in the maillog file:
>
> Dec 28 10:08:10 mail03 postfix/qmgr[17701]:
56CF71AF421: removed
>
> Is there any else that could help me find a solution to
it?
>   

Where is amavisd-maia running?  both machines? somewhere
else?

Do both of these machines then forward to somewhere else?

look for any messages about connections being refused... 
Make sure you
have enough amavisd-maia children for your postfix servers -
if they are
sharing, one could be starving the other.

_______________________________________________
Maia-users mailing list
Maia-usersrenaissoft.com
http://www.renaissoft.com/mailman/listinfo/maia-users