Dispatcher +SMTP service fail continuously

Question

# ../../imsimta version
iPlanet Messaging Server 5.2 Patch 2 (built Jul 14 2004)
libimta.so 5.2 Patch 2 (built 19:30:12, Jul 14 2004)
SunOS mail 5.10 Generic_118833-36 sun4u sparc
I have the dispatcher failing continuously so that I had to restart it every 30 minutes.
Here is the dispatcher.cnf:
# grep -v "^\!" dispatcher.cnf
MIN_PROCS=1
MAX_PROCS=40
MIN_CONNS=30
MAX_CONNS=50
MAX_SHUTDOWN=2
MAX_LIFE_TIME=86400
MAX_LIFE_CONNS=10000
MAX_IDLE_TIME=600
HISTORICAL_TIME=0
[SERVICE=SMTP]
PORT=25
IMAGE=/usr/iplanet/server5/bin/msg/imta/bin/tcp_smtp_server
LOGFILE=IMTA_LOG:tcp_smtp_server.log
STACKSIZE=2048000
[SERVICE=SMTP_SUBMIT]
PORT=587
IMAGE=/usr/iplanet/server5/bin/msg/imta/bin/tcp_smtp_server
LOGFILE=IMTA_LOG:tcp_smtp_server.log
PARAMETER=CHANNEL=tcp_submit
STACKSIZE=2048000
Here is also a message the sometimes appear in the dispatcher log file:
# more dispatcher.log-0K8O00001HWOBP
16:34:55.09 ( 3): app_create_proc(SMTP): too many processes for this
service (39 Current + 1 Pending + 0 Shutting down >= 40
Maximum)
The machine has 4x1.8 GHz CPU + 16GB Memory + 1 Gbps interface
Could somebody help me to pinpoint the problem and tune up the system?
Thanks,

Answer 1

msragab wrote:
I have many DNS servers in /etc/resolv.conf. and they are all working (I tested that using nslookup). In this case it isn't enough that they are working, they need to be responding quickly so that connections don't queue up. Also you may want to try restarting the nscd daemon on the Solaris server to see whether this helps.
One thing I have discovered is during the hang time of the dispatcher, "./imsimta dispatcher_stats_tty"shows many connections (more than normal) from the anti-spam gateway machine. But I think this is normal as the MTA at the email server is not responding ( I expect it is not finishing work also).That would make sense. Have you included the IP/hostname of this system in your /etc/hosts file? This will help to reduce the impact of lookups for those connections as Messaging Server caches the contents of /etc/hosts on startup for name resolution.
This is a dedicated email security device from Symantec.
Does this present any clues?Unfortunately not. The usual question that is asked by support in cases like this is what changed in your environment around the time of this problem e.g. network/application/patches/server changes including rebooting/system load and so on.
Also is the server itself running out of resources e.g. high CPU load in vmstat, high block-on-IO in iostat and so forth.
Also, regarding the command "./imsimta dispatcher_stats_tty", how do I understand its output? Are there documentation for it?This command is used more for troubleshooting/support purposes so there is no external documentation.
Regards,
Shane.