Server-Agent Connection mode : Transient

sharas · Post by **sharas** » 14 Dec 2007 8:35

some jobs have gone to unknown status ,when I did ag_menu its was showing like this. what can we do to get the
Server-Agent Connection mode up

Server-Agent Connection mode : Transient
Unix Ping to Server Platform : Succeeded
Agent Ping to Control-M/Server : Succeeded

thnks

philmalmaison · Post by **philmalmaison** » 14 Dec 2007 9:44

your version seems to be 6.1 or less no ?
could you send the followings:
agent os.dat / agent config.dat

look at services mecanism (/etc/services /etc/inetd.conf)
Your agent services arre under inetd process managment...

last have a look to the port numbers (S to A A to S)

regards
Philmalmaison

sharas · Post by **sharas** » 14 Dec 2007 10:42

my agent is 6.2.01.

and i see there are many p_ctmag processes created and the agent is hung up, but when i ping it from server i see it alive. and if try to shut the agent i am unable to do that. and all the jobs turned to white, ie unknown status.

philmalmaison · Post by **philmalmaison** » 14 Dec 2007 10:51

you must have two process online :
p_ctmag
p_ctmat (tracker)

look at the files locate in ctm/proclog of the agent AT*.log , verify if there are no semaphores problems ... or other system pb...

regards
Philmalmaison

sharas · Post by **sharas** » 14 Dec 2007 1:06

actually I have no idea of semaphores and how to troubleshoot them, can you please go a litlle deeper and explain me.

Thank you very much for your inerest in answering me.

philmalmaison · Post by **philmalmaison** » 14 Dec 2007 4:34

this is a system problem, semaphores are about memory usage...
you should set it up by two ways in Sol10
/etc/project
/etc/system
recommanded is :
set shmsys:shminfo_shmmax=4294967295
set shmsys:shminfo_shmmni=1024
set shmsys:shminfo_semmns=16384

modifying /etc/system need reboot to take effect

regards
Philmalmaison

Post by **fyot** » 15 Dec 2007 9:03

So

What's your fixpack?
Do you have big sysout as > to 10 MB ?
Are you working with Solaris 2.8 ?

If you are concerned by that.
I had same issue, and here is the solution.
A case is open to BMC, and they don't have any solutions for the moment.

Reason
Jobs are executed correctly, but when the sysout is analysed by the tracker, there is some internal problems, and the tracker run on itself.
If you run a prstat or top utilities, you could see tracker using a lot of memory and doesn't free it.
So In fact your job wait in executing mode the total return from the tracker, and it could wait for a long time.

If you try to stop your agent, p_ctmag could be stopped, but not the p_ctmat. To start again correctly your agent you must kill p_ctmat process.

Solution :

Install from scratch your Unix Agent and deploy only the Fixpack 1.
It should be run correctly

Of course don't forget too, to apply system requirement explained by philmalmaison, before.

sharas · Post by **sharas** » 18 Dec 2007 1:12

Hi Philmalmaison,

can you please tell me what are these for

set shmsys:shminfo_shmmax=4294967295
set shmsys:shminfo_shmmni=1024
set shmsys:shminfo_semmns=16384

thanks,

philmalmaison · Post by **philmalmaison** » 18 Dec 2007 1:38

memory size parameters necessary to run agent process

set shmsys:shminfo_shmmax=4294967295
Maximum shared memory segment size

set shmsys:shminfo_shmmni=1024
Number of shared memory identifiers

set shmsys:shminfo_semmns=16384
Number of semaphores in the system

regards,
Philmalmaison