how to start sendmail in hp-ux

[senthil_kumar_1]

Hi

I want to find whether sendmail is running or not, so i issued following command.

root@lghp239:/root > ps -ef | grep sendmail
root 732 1 0 Oct 31 ? 6:51 sendmail: rejecting connections on port 25: load average: 62


what it "sendmail:rejecting connections" means.

how to start sendmail.

[Avinash20]

Have a look at your /etc/mail/sendmail.cf file and search for 'Load'.

You will see a documentation section in the file titled "Refuse-Connections Load Average (option RefuseLA):".

To see how this is configured on your system now, search for 'RefuseLA' in the sendmail.cf file. If the line looks like:

#O RefuseLA=12

then you are probably using a built-in default value.

If the line looks like:

O RefuseLA=15

Then the load average at which sendmail will start refusing connections is 15. The value could be any number you choose.

If having sendmail refusing connections is a real problem, you probably really need to look into why the load average was so high to begin with.

[Avinash20]

In your case the load average seems to be high
Reduce the load average on the server and you won't get this rejected ..

[Jeff Schussele]

I agree with Ravinish.
When it's refusing connections do a

ps -ef | grep sendmail

to get the PID - 732 - in this case.

Then do

ps -ef | grep 732 | wc -l to see how many children the parent - 732 - has spawned.
If that equals/exceeds the RefuseLA value that would explain the rejecting state.

HTH,
Jeff

[senthil_kumar_1]

Hi

I edited the file "/etc/mail/sendmail.cf"

o/p: #O RefuseLA=12

# ps -ef | grep 732 | wc -l

o/p: 3

how to solve this problem.

[Jeff Schussele · Accepted solution]

OK - I think then your next step would be to run up top or glance to see whether you've exhausted memory and/or swap. Could even be that you've reached the NPROC max & can't fork anything until the PID count falls.

Dos a ps -ef | wc -l and compare that to the NPROC value which you can check with

kmtune -q nproc

But of course, IF you've hit the NPROC limit you may not be able to run anything until you kill some PIDs.

Check your syslog - anytime your hit a kernel parm limit it should get logged. You should also check your mail log - it could/should tell you exactly what's going on.

HTH,
Jeff

[Avinash20]

Please post the following output here.

# top
# vmstat 2 2
# swapinfo -tam
# sar -Mu 2 2
# sar -d 2 2
# ipca -ma