Title: Re: [Sipp-users] sipp defunct processes
Vlad,
 
on defunct:
-----------
you're right, it's the way daemons and backgrounding is done, and solves the "defunct" issue. So, the ideal bug fix.
 
I have some problems though with the final result we get, but those are not breaking to use this solution now:
 
1. to execute a (small) unix program, we will use "fork - fork - system", that gives three process creations instead of one (but we already had two in the code;-), it's certainly not optimal/efficient, but computers are fast with a lot of resources these days.
 
2. making an orphan of the final child process gives them ppid=1, making tools like pstree and pkill fail to identify the correct child processes. When running multiple instances of sipp (as I do), it might give problems to handle all processes hanging around (cleanup when things don't go like you expected).
 
Olivier,
 
on IP impairments:
-----------------
 
I noticed you proposing NISTNET as impairment generator on linux, but if my info is correct, most people nowadays have switched to NETEM, which is integrated in the standard linux kernel (you might have to activate it when compiling). Netem is only one of the tools available in a more general scheme, which also support chaining with packet classifiers, queues and shapers. Very powerfull, I only used the basic impairments until now (per direction drop, duplicate, delay/jitter).
 
Best regards,
 
  MarcVD

From: Vlad Troyanker [mailto:[EMAIL PROTECTED]
Sent: 24 October 2006 19:18
To: [EMAIL PROTECTED]
Cc: [email protected]
Subject: RE: [Sipp-users] sipp defunct processes

Marc,

 

The fix works by spawning a child then the child immediately spawning another child (ch#2) that ch#2 will do the actual work. Parent waits for ch#1 which returns immediately so no blocking occurs there. After ch#2’s parent exits its new parent becomes system ‘init’ process which will properly reap it once completed.

 

The above might sound confusing but that is how it works (dark corners of UNIX legacy). I faced the same problem and made this fix for myself a few months ago and have been happy with it ever since. Several people are using sipp here. I had a real trouble submitting a message to the forum because it uses very strong anti-spam (Sender Call Out).

 

Best Regards,

Vlad

 


From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED]
Sent: Tuesday, October 24, 2006 12:50 AM
To: [email protected]
Subject: Re: [Sipp-users] sipp defunct processes

 

>Marc,
>can you confirm that this is fixed for you too?
>
>http://sipp.sourceforge.net/snapshots/sipp.2006-10-23.tar.gz
>
>Thanks,
>Olivier.

Olivier,

As I won't have time until somewhere in November to try this out, you'll have to wait a bit for my feedback.
But if I read the code example in the mail correctly, I have the impression that the parent waits for the child immediately, which will block it until the child has finished. It would be sufficient for what I plan now (send out an snmp trap), but not if the child gets some more work to do. Of course, you don't have any defuncts this way.

But, as said before, I'll test it somewhere in November,

  Best regards,

   MarcVD

(-: from Marc VAN DIEST (BELGACOM ANS/NTA) +32 2 244 5078 ;-)

**** DISCLAIMER **** http://www.belgacom.be/maildisclaimer
**** DISCLAIMER **** http://www.belgacom.be/maildisclaimer
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Sipp-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sipp-users

Reply via email to