from what I understood and the problem we were having...yes.

yes, they will both work, however, the phantom processes that are
    forked off, when they end, their PID are not released back into
    the PID pool until the parent phantom dies.

If the parent phantom only forks off a few phantoms it may not be noticable,
  however, when the parent phantom is forking off 20-30-40 child processes
  every minute, after about 6-7 hours, this became a major problem having
  all those zombie processes hanging around.

It's also possible this might just be a unix thing, (zombies).

It's very possible that eventually the zombie's may be released over some
period of time, and with few child processes, it never will become a problem.
However, we were using it in conjunction with a webserver, and each cgi request
would filter through our phantom and fire off the appropriate program to
dynamically create the HTML code that would then get fed back to the cgi. 

If I'm mistaken, which is very possible :) and our situation was unique, I'm
sure the correct information will be proudly posted (which is always a good 
thing,
the truth must be told!).

> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] Behalf Of Bill Haskett
> Sent: Friday, May 19, 2006 1:05 PM
> To: [email protected]
> Subject: RE: [U2] Phantom job started from another phantom job
> 
> 
> George:
> 
> Are you saying one cannot have a single service manager 
> program running once
> as a phantom, like cron?  So, this doesn't work:
> 
> PHANTOM RUN MYCRON
> 
> MYCRON
> 001 LOOP
> 002    ...Read if MyCron should be stopped
> 003    ...Stop MyCron if it should be stopped
> 004    ...Read jobs to run
> 005    IF JobToRun THEN
> 006       PHANTOM RUN JOBDEFINITION
> 007    END
> 008    SLEEP 60
> 009 REPEAT
> 
> ...but this does:
> 
> PHANTOM RUN MYCRON
> 
> PHANTOM.MYCRON
> 001 PA
> 002 PHANTOM RUN MYCRON
> 
> MYCRON
> 001 LOOP
> 002    ...Read if MyCron should be stopped
> 003    IF StopMyCron THEN STOP
> 004    ...Read jobs to run
> 005    IF JobToRun THEN
> 006       PHANTOM RUN JOBDEFINITION
> 007       CHAIN \PHANTOM.MYCRON\
> 007    END
> 008    SLEEP 60
> 009 REPEAT
> 
> Do I understand this correctly?
> 
> Bill
>  
> 
> > -----Original Message-----
> > From: [EMAIL PROTECTED] 
> > [mailto:[EMAIL PROTECTED] On Behalf Of 
> George Gallen
> > Sent: Friday, May 19, 2006 8:16 AM
> > To: [email protected]
> > Subject: RE: [U2] Phantom job started from another phantom job
> > 
> > After sleeping, instead of the process repeats itself, what 
> > you need to do is CHAIN"" to itself, this will end that 
> > phantom, which will release all it's child processes from 
> > their zombie status. 
> > 
> > If you continue to build up zombies, without releaseing them, 
> > you will eventually get the dreaded "no more processes available"
> > message, at which point, UV comes to grinding crash halt, 
> > until you kill that one repeating phantom, then magically, 
> > all the processes become available again, and UV picks itself 
> > up and continues on with life.
> > 
> > The only problem with this method, is you will build up a 
> > crap load of very small files in the &PHANTOMS& file. We have 
> > a process which every few days deletes old fileid (using the 
> > ls to give us the date the file was created, and made our 
> > &PHANTOMS& file a type 19)
> > 
> > Oh, and YEA, looks like the reply is back to the way it was.
> > 
> > George
> > 
> > > -----Original Message-----
> > > From: [EMAIL PROTECTED]
> > > [mailto:[EMAIL PROTECTED] Behalf Of 
> Scott, Dave
> > > Sent: Friday, May 19, 2006 10:56 AM
> > > To: [email protected]
> > > Subject: [U2] Phantom job started from another phantom job
> > > 
> > > 
> > > A colleague of mine ask me to post this on his behalf.
> > > 
> > > I have a Phantom job that runs continuously. The job performs a 
> > > select, executes some processing on the records returned from the 
> > > select, and then goes to sleep for a few minutes. After 
> > sleeping, the 
> > > process repeats itself. I recently added an additional step 
> > after the 
> > > processing. This step starts another Phantom process which 
> > launches a 
> > > program that does a select and processes the results of the 
> > select and 
> > > then stops (using CHAIN "OFF").
> > > 
> > > The problem is that our system admins have noticed that 
> the "child" 
> > > Phantom processes launched by the parent Phantom process 
> go into a 
> > > state of "defunct"
> > > from a Unix process perspective.
> > > 
> > > $ ps -ef | grep defunct
> > >  fhooven 23180   986 18 16:33:12 ?         0:00 <defunct>
> > >   dscott 23302 23199  1 16:33:32 pts/tW    0:00 grep defunct
> > > 
> > > The child Phantom process runs successfully and the 
> > "defunct" process 
> > > eventually goes away.
> > > 
> > > Any idea what causes this?
> > > 
> > > System is Universe 10.0 running on HPUX B.11.11 A
> > > 
> > > Thanks for you help.
> > > 
> > > ...Frank Hooven
> > > -------
> > > u2-users mailing list
> > > [email protected]
> > > To unsubscribe please visit http://listserver.u2ug.org/
> > -------
> > u2-users mailing list
> > [email protected]
> > To unsubscribe please visit http://listserver.u2ug.org/
> -------
> u2-users mailing list
> [email protected]
> To unsubscribe please visit http://listserver.u2ug.org/
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

Reply via email to