Re: Too slow ?

[Matti Aarnio <mea@nic.funet.fi>]

On Wed, Mar 12, 2003 at 08:25:26PM +0100, Tomasz Nowak wrote:
> Hi,
>   My problem is:
>     1047492376 85723-12310 2212 12 ok3 local/...
> 
>   I have too much lines like this in scheduler.perflog.
> 
>   How can I find a bottle-neck ? Any hints ?

  2212 seconds from message arrival to its routing ?
  (Key is given in  STATISTICS LOG FORMAT  of  scheduler(8) man-page.)

  That is definitely quite a lot..
  Do you have lots of files in $POSTOFFICE/router/  directory ?
  ("mailq -ss"  will tell.)


  This format is no longer giving any _easy_ correlation to
  syslogged entries.  Look for ' S85723' string in your
  syslogged data.  (E.g. add ' S' to i-node number from the
  second field of that log format up to '-' character.

  It is possible to convert the first two fields to spoolids,
  that are logged into syslog.
 ...
  I did commit a tool to CVS:
    scheduler/statuslog-ids-to-spoolids.pl
  you may, perhaps, need to fix reference to which perl you have..
  Feed into it the  scheduler.perflog  (extracts), then you get
  out spoolids.

  In   syslogged  data, you can find items logged by  router,
  as well as transport-agents.

  There are   delay=   and   xdelay=  time stampts.

  If you see  router[...]  produced entries where  delay=  is
  high, then there is definitely something odd in message feeding
  to routing.

 ...

  Are there perhaps alternate 'router' directories ?
  (the "mailq -ss" does not show all of those..)

  In that case, older queue scanners can't guarantee, when
  a "lower-priority" queue is picked next time around.
  The  smtpserver  does look for  Precedence:  header,
  and knows of "high", "bulk", "junk", "normal", of which
  "bulk", and "junk" are lower than "normal/high".


  Lattest code tries to allocate one router process to
  each alternate (lower priority) router-spool directory,
  but quick reading gives me a reason to wonder if that
  is really correct in all cases..  (if router's -n
  option is lower than the number of existing   ROUTERDIRS
  pointed real directories, for example..)


  If your ZMailer version is old enough (mere year) so that
  there are no  $POSTOFFICE/.*.notify  sockets, then the system
  will have to search for directories for jobs.
  (This needs  zmailer.conf  variables pointing to those sockets..)

  Even a new one _might_ sometimes miss new messages, and will
  then use passive scanner to spot them.  (That runs around every
  10 seconds, or so.)


  Does your (presuming _newish_ routers)    router   -log have
  "ROUTER CHILD" ...  strings ?  Indicating something wrong ?
  ("STARTING", and "ROUTER TERMINATING" -> "EXIT 1"  are ok.
   .. that exit-code is a bit illogical.  I will change it to 0...)


> Pozdrawiam.
> Tomasz Nowak     TRIGER - Systemy Komputerowe   http://www.triger.com.pl

-- 
/Matti Aarnio	<mea@nic.funet.fi>
-
To unsubscribe from this list: send the line "unsubscribe zmailer" in
the body of a message to majordomo@nic.funet.fi