MailScanner running very slow

[Eduardo Casarero]

2008/8/13 Nasser Al-Zawawi <nassera at alz-inc.com>

>  Hi,
>
> I have a Dell PowerEdge server 2950 with 2 Dual Core Xeon processors 3.0GHz
> with 2GB RAM running RedHat ES 4.  The MTA is sendmail and I am using the
> latest MailScanner version (4.70.7-1).  The system processes about 2,000
> emails per day and about half of it are spam.  See the following:
>
> #---------------
>
> MailScanner Status:
>
>       1920 messages Scanned by MailScanner
>
>       73657132 Total Bytes
>
>       1122 Spam messages detected by MailScanner
>
>       6 Viruses found by MailScanner
>
>       935 Messages delivered by MailScanner
>
> #---------------
>
>
>
> The mail delivery is very slow.  Some time it takes 2 hrs before it gets
> delivered.  I tried the tips under
> http://wiki.mailscanner.info/doku.php?id=documentation:tweaking:some_things_to_try_if_your_incoming_queue_is_running_slowbut they did not help.
> The mail keeps accumulating under /var/spool/mqueue.in while all the
> MailScanner processors are running at 100% CPU.  Then all the sudden
> (after 1-2 hours) it starts delivering the emails and the /var/spool/
> mqueue.in files go down to 0.
>
> I monitored the system today; the system stopped delivering emails between
> 9:12AM till 11:35AM.  During that time, top showed something like this:
>
> #----------------
>
> top - 10:46:26 up 3 days,  7:42,  1 user,  load average: 5.81, 6.09, 5.48
>
> Tasks: 233 total,   5 running, 226 sleeping,   0 stopped,   2 zombie
>
> Cpu(s): 59.0% us,  1.2% sy,  0.0% ni, 39.5% id,  0.2% wa,  0.0% hi,  0.0%
> si
>
> Mem:   2074600k total,  1735692k used,   338908k free,    22524k buffers
>
> Swap:  2097144k total,   356516k used,  1740628k free,   532576k cached
>
>
>
>   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>
>  2752 root      25   0 85572  75m 2964 R  100  3.7  35:54.85 MailScanner
>
>  2880 root      25   0 85572  75m 2956 R  100  3.7  35:50.35 MailScanner
>
>  3024 root      25   0 85572  75m 2956 R  100  3.7  35:41.62 MailScanner
>
>  2949 root      25   0 85572  75m 2956 R  100  3.7  35:47.07 MailScanner
>
> 11393 apache    15   0 26360 9736 3648 S    5  0.5   0:57.14 httpd
>
> 11394 apache    15   0 26576 9936 3616 S    5  0.5   0:58.83 httpd
>
> 26598 root      16   0  4760 2568 1648 S    1  0.1   0:00.20 ssh
>
> #----------------
>
>
>
> and the /var/spool/mqueue.in continued to accumulate files till it reached
> 489 messages, then within 5 minutes it delivered all the messages!
>
>
>
> This cycles continues all day long, where the emails are withheld for 1 to
> 2 hours and then the get delivered in one big burst.
>
>
>
> Here are snippets of the log for that period with "Log Speed = yes":
>
>
>
> grep ": Batch" /var/log/maillog
>
> #-----------
>
> Aug 13 09:12:16 samba MailScanner[5852]: Batch completed at 1844 bytes per
> second (39583 / 21)
>
> Aug 13 09:12:16 samba MailScanner[5852]: Batch (2 messages) processed in
> 21.46 seconds
>
> Aug 13 09:12:42 samba MailScanner[6013]: Batch completed at 3752 bytes per
> second (9167 / 2)
>
> …
>
> Aug 13 11:28:55 samba MailScanner[2949]: Batch (10 messages) processed in
> 24.54 seconds
>
> Aug 13 11:28:55 samba MailScanner[3024]: Batch completed at 657 bytes per
> second (3825 / 5)
>
> Aug 13 11:28:55 samba MailScanner[3024]: Batch (1 message) processed in
> 5.82 seconds
>
> #-----------
>
> I would appreciate any clues on resolving this issue.
>
>
>
Are you using a local caching dns? dns are critical, remove rbl checking to
see if speed increases.

Do you use clamavmodule? switch to clamd, are you using tmpfs for
/var/spool/MailScanner/incomming?

delete spamassassin cache and the bayes DB. sometimes they get corrupted and
mess things up.

Regards,

Eduardo.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mailscanner.info/pipermail/mailscanner/attachments/20080813/7a09e7f5/attachment.html