Kelly,

Here's  something  very wrong with your setup, to start off: you don't
allow  <>  senders. A cursory look at the archives would have told you
NEVER  to  enable that option. And yes, depending on bounce volume, it
can  cause  big  ol'  problems,  though there are more likely culprits
here.

> Ok, I've been using Imail for probably 5 years for our hosting company -- we
> have about 2000 virtual hosts & ~7000 users.

Not  really  relevant  metrics.  You  need  to  know  bytes in/out and
messages  in/out: that's your usage for SMTP. IWEBMSG usage is another
story,  and  again  your number of *potential* users is irrelevant vs.
your  actual  number  of  bytes  transferred, and number of concurrent
sessions, in practice.

> About once a day or so, we have to restart the SMTP server

Okay,  question  #1:  Is  it chewing up the CPU at that moment (sounds
like it's not, from what you said below)? Or is it just not responding
to new sessions, and hanging in-progress ones? How many spool files do
you  orphan, and with what distribution of extensions and D file sizes
(start  running averages now)? Any similarity in the recipient Q files
at  the  time  the  service  hangs  (i.e. corrupt mailboxes)? How many
SMTP32s are usually running at the time?

> Today, smtp was dying more frequent than ever!  Web messaging seems much
> more slow, and IMAP is frequently timing out!!

Like  John  said:  NIC,  NIC,  NIC...this  is  one of the first things
people'd  swap  in  a workstation. Can't imagine why more people don't
put  this  on  the  top  of  the  server  troubleshooting  shelf.  And
switchport errors, patch errors, etc.

> I have dropped two name servers listed in the lookup box to one.

This  seems  to be a common suggestion from support that has picked up
some  steam  recently,  but  I do not believe that it is grounded in a
real  problem  with  IMail. However, it is not a bad idea to rotate in
your  DNS  servers to see if one or more may be broken or slow, though
simply  leaving  the  first  server  in  there arbitrarily is not very
scientific.

> I  have  moved  the  spool directory to a different drive (having to
> break a mirror to try this).

Geez,  you could have just bought a micro-cheap IDE drive for testing.
:)

> I've  set the log file to the log server so I can watch smtp traffic
> real time. Nothing seems abnormal.

So  you're  not  seeing  anything  in  the  real-time logs (unfinished
processes),  nor  anything  consistent in the spool at the moment that
the  server hangs? How about PerfMon baselining of CPU and network I/O
to compare to SMTP restart times?

-Sandy


To Unsubscribe: http://www.ipswitch.com/support/mailing-lists.html
List Archive: http://www.mail-archive.com/imail_forum%40list.ipswitch.com/
Knowledge Base/FAQ: http://www.ipswitch.com/support/IMail/

Reply via email to