Process Numbre with NetScape Proxy 3.5 and Solaris

[VinZ]

Hello,

I have a lot of NetScape 3.5 proxies instances on SPARC/Sun Solaris.
In my /opt/proxy/proxy-instance/config/magnus.conf i define MaxProcs 64.
But i have always more than 64 process launched in memory. Randomly between 65 and 70.
A lot of this child process are alone (except the father of course).
Exemple :
PID Father Process name
1999 1 ./ns-proxy -d /opt/proxy/... (Father process)
2000 1999 ./ns-proxy -d /opt/proxy/...
2001 1999 ./ns-proxy -d /opt/proxy/...
....
2066 1 ./ns-proxy -d /opt/proxy/... (Whithout father)
2067 1 ./ns-proxy -d /opt/proxy/...

I restart the instance and i kill the process not killed by the ./stop and ./start procedure, but it's the same situation.

Can U help me please ?

Thanks in advance.

 

[jad]

[tt]MaxProcs

The MaxProcs directive sets the maximum number of processes the server can have active. The MaxProcs number is the number of processes that the proxy is to constantly keep active. When you change MaxProcs, you must do a hard restart for the change to take effect. Choose a number that is appropriate for the volume of access you expect for the proxy server. If this number is too small, clients will experience delays. If the number is too large, you might waste resources that other programs could use. Syntax

MaxProcs number

number is a whole number between 1 and the size of your system's process table. Default

MaxProcs 50
[/tt]

in otherwords MaxProcs is the minimum and maximum number of processes ... nice ehh?

you could always try a nice free proxy called squid

http://www.squid-cache.org

... we've never had problems with it on our solaris boxes

 

[VinZ]

Thanks, but i know.. I have a lot of users behind this proxy instance (About 6,000) and i had define my MaxProcs value in consequence.
This problem not affect proxies performance, but the number of process is over 64 (Values define in my magnus.conf). It dosen't matter, but i check the number of process with TNG and if i have more than 64 processes TNG send a critical alarm.

 

[jad]

well it might well be 64 processes which themselves have up to 5 child processes ... that'd make up to 5 times the amount of processes ...

if you open up netscape for instance you don't get 1 process for netscape, it has about 4 processes just to get one window. (after that it adds just one process for each window, but thats not the point)

it may well have to do dns lookups which might well require it to spawn off new processes to different dn servers.

if it is doing any number of these things then you're a bit stuck either way.

you can set something in the NS cache to kill off child processes after a set time ... it might be worth while setting that to cut the number of child processes down.