Mail server saturation

7 posts / 0 new
Topic locked
Last post
#1 Sun, 12/25/2011 - 14:37
ricks

Mail server saturation

Hi,

I use webmin to host some of my website. When all person that use our server for their mail (10 person use simultaneous, trought iMap and Mozilla ThunderBird, some of them use same account simultaneously).

When all person use their Mozilla, we have a problem, server don't reply and we cannot read new mail or send an email correctly (the email was sended, but isn't copied on the server, Mozilla said that he can't copy the mail to the outbound folder). And if i try to read mail in the same time trought Virtualmin WEBMAIL, it was ver very very slow...

Have you an idea to solve my problem?

I think that we have limited process that run IMAP, can i unlimit them?

Please note that we run Ubuntu on an OpenVZ.

Regards, Eric

Sun, 12/25/2011 - 19:07
andreychek

Howdy,

Well, 10 users accessing email isn't actually very much -- servers can typically handle hundreds of simultaneous email users.

How much RAM does your server have? You can determine that by running "free -m".

Also, what is the output of this command:

cat /proc/user_beancounters

Tue, 12/27/2011 - 01:27
ricks

I think that i made a mistake, each user have 7-10 mail accounts (5 of them are common to all users, and used permanently trought thunderbird).

In the morning, before everyone start to working, cat says :

[quote] cat /proc/user_beancounters Version: 2.5 uid resource held maxheld barrier limit failcnt 110: kmemsize 50933039 85699932 9223372036854775807 9223372036854775807 0 lockedpages 0 10 3670016 3670016 0 privvmpages 831764 1230789 7340032 7352532 0 shmpages 15782 30914 9223372036854775807 9223372036854775807 0 dummy 0 0 0 0 0 numproc 394 637 1024 1024 0 physpages 669695 1012082 0 9223372036854775807 0 vmguarpages 0 0 7340032 9223372036854775807 0 oomguarpages 669703 1012082 7340032 9223372036854775807 0 numtcpsock 49 420 9223372036854775807 9223372036854775807 0 numflock 14 120 9223372036854775807 9223372036854775807 0 numpty 1 1 255 255 0 numsiginfo 0 46 1024 1024 0 tcpsndbuf 3860184 29105104 9223372036854775807 9223372036854775807 0 tcprcvbuf 802816 4104824 9223372036854775807 9223372036854775807 0 othersockbuf 1338720 1863584 9223372036854775807 9223372036854775807 0 dgramrcvbuf 0 11032 9223372036854775807 9223372036854775807 0 numothersock 742 951 9223372036854775807 9223372036854775807 0 dcachesize 2461920 3340248 9223372036854775807 9223372036854775807 0 numfile 14479 19920 9223372036854775807 9223372036854775807 0 dummy 0 0 0 0 0 dummy 0 0 0 0 0 dummy 0 0 0 0 0 numiptent 10 10 9223372036854775807 9223372036854775807 0 [/quote]

Tue, 12/27/2011 - 12:35
ricks

You can see on this link more cat, when we are in charge and when the problem come.

Tue, 12/27/2011 - 12:36
ricks

You can see on this link more cat, when we are in charge and when the problem come.

Tue, 12/27/2011 - 14:48
andreychek

Well, that part looks normal. I had wondered if maybe you were running into any OpenVZ resource limits, as that's a common source of problems.

How about RAM -- what does "free -m" look like on your system?

I've seen systems with hundreds of domains and thousands of email accounts run perfectly normal on a default installation without having to modify any resource settings for Dovecot (IMAP/POP) -- so I'm looking to see if there's any other issues going on before we look into Dovecot-specific issues.

-Eric

Tue, 12/27/2011 - 17:05
ricks

Hi,

Here the extrat of free -m command...

Thanks for your help

Topic locked