>10k emails in active queue... :(

[Predatorz]

Hi, I am using zimbra OSS v4.5.5.

Over the weekend, our mail server received >10k Emails, got choked in the active queue thus causing problems in receiving/sending email. we have since rebooted the server and after fixing the clamd not able to start up proble, the mail server is back online.
We have moved the >10k email to the Hold Queue.

How do we increase the active queue size to accomodate >10k of emails?
And is there any good way for us to purge all the emails to all the mail recipients while allowing the mail server to continue to send & receive emails?


Thanks for any help/suggestions rendered.

[jholder]

Hi, I am using zimbra OSS v4.5.5.

Over the weekend, our mail server received >10k Emails, got choked in the active queue thus causing problems in receiving/sending email. we have since rebooted the server and after fixing the clamd not able to start up proble, the mail server is back online.
We have moved the >10k email to the Hold Queue.

How do we increase the active queue size to accomodate >10k of emails?
And is there any good way for us to purge all the emails to all the mail recipients while allowing the mail server to continue to send & receive emails?


Thanks for any help/suggestions rendered.

Hi Predatorz,
Welcome to the forums, and I'm sorry that you're having trouble.
Zimbra should have been able to handle this. If you've fixed the clam av, problem, then I would suggest requeing the e-mail messages and then running postqueue -f.

You can also turn off clamav for the time being, and deliver the messages. We should be releasing 4.5.7 which fixes this issue, very soon.

Sincerely,
john

[mmorse]

I'm supprised their in the active queue, the clamav problem usually has them in the deferred. Once you fix your problem, they won't (unless forced) all try again at the exact same time. The active queue size is limited on purpose, because it deals with current memory.

You said that their now in the hold-just release them-and if need be do it in groups. Just re-queuing as apposed to then flushing after you release, may reduce the strain a bit. (If this was v5beta-with posfix 2.4, you could also do postqueue -i queue_id, to just deal with the hold queue and not affect any mail in the other queues, but you did say v4.5.5 which has an older version of postfix.)

Though postsuper -r ALL hold (hold can = any queue_id) should work. However, I always avoid 'postsuper' because -r reque is easily confused with -d delete; you could accidentally think it was 'remove' and 'delay' instead. And -h and -H have totally different purposes -one holds the other releases. And -H ALL hold acts different than -r (-H keeps the max lifetime running!) The confusion happens more than you think, which is why I pull up a man page first when using postsuper.

John is also exactly right when he says "Zimbra should have been able to handle this. If you've fixed the clam av, problem, then I would suggest requeing the e-mail messages and then running postqueue -f."

Deferred can scale quite bigger...If you still want to increase active - ram issues asside:
qmgr_message_active_limit which is the max number of messages the active queue can hold.
So it would be like: postconf -e 'qmgr_message_active_limit = 20000'

While your at it, be aware that the sum total of the recipeients in the active queue won't exceed qmgr_message_recipient_limit/default_recipient_limit

(Because I know you'll come across it...smtpd_recipient_limit refers to the max number of recipients per email, and it applies to all queues. And if you mess with that be advised of your smtpd_recipient_overshoot_limit)

[Predatorz]

Am i right to say that postqueue -f will flush all emails in all queues(Deferred, Incoming, Active, Hold & Corrupt?)

Thanks again!

[jholder]

Postqueue should flush everything. If things still hang around, please post your /var/log/zimbra.log

[plumsicle]

Hello,

I don't get 10k emails in the active queue but I can have anywhere between 1k and 2k. The problem is, is that the emails are being delivered very slowly as the antispam-virus is scanning all the emails and takes a very long time. Many spam emails are being sent to users that don't even exist in my domain yet the amavsid is still scanning them and it's bogging everything down. I've looked on various threads and I have tried to resolve the issues but to no avail. I'm running the NE version 5.0 and I've had trouble finding information on this Clam AV problem to see if I have it. We've been running through zimbra for about a year now and it only started to occur around November during the time we were running 4.5.7 and upgrading did not help as much as I had hoped.

Thanks in advance,
Jason.

[mmorse]

You may wish to change the entry in /opt/zimbra/conf/zmmta.cf for smtpd_reject_unlisted_recipient from 'no' to 'yes', save the file, then as the zimbra user run 'postfix reload'.

[p24t]

What is the load on the server like? Something like that sounds more to be a capacity issue. Number of users/hardware spec?

[plumsicle]

I've already changed that setting in zmmta.cf as per another thread. The issues still continued where emails go up to about 1000 and takes about an hour for me to send an email within the domain and even to myself.

Running a quad-core 2.8 Xeon processor with 4GB of RAM.

We have approximately 120 users.

Thanks,
Jason.

[plumsicle]

Another thing, would Postini have any impact on this situation? We are running that as our spam filter before hitting the zimbra server.

Thanks again,
Jason.