We have a database/agent purchased from Lotus FSS a couple of years ago. This app is designed to read through all of the mail files on a server and purge unfiled messages from (basically) the inbox and sent view that haven’t been accessed in over ‘x’ days.
On our first Release 6 server, we have about 10 test users. To process this many people should take about 15 minutes. Apparently, there is something going wrong on that server, as the most recent run started on 2/22/03 at 3:30AM, and it hasn’t completed yet (as of 2/25/03 at 6:00AM).
There are some obvious problems going on elsewhere, but what has me most concerned is that the agent is still running. Our nighttime max execution time is set to 300 minutes, to accommodate this agent on our larger mail servers. Daytime execution is set at 15 minutes. Over the course of 3 days, Amgr should have killed this agent because of the extensive runtime, but hasn’t.
If I need to open a support incident on this, let me know.
Subject: Agent Manager does not honor max execution time
“Troubleshooting agents in r5 and R6” article might be helpful with suggestions of the server console to try. “Decoding agents in R6” explains the new command that will allow you to kill agents from the server console.
You will find both in agent FAQ (faq#0) or in article archives of this site under my name as author.
the task list on the admin client clearly indicated that this agent was running,
the agent did in fact perform everything it was supposed to do,
when issuing this command the server returned that “AMgr: Agent ‘Agent Name’ in ‘dbname.nsf’ is not running.” When I issue the ‘q’ at the server console, 2 of my 3 agent executives shut down, while one hangs. and,
this agent which runs every 30 minutes, shows the last time as hours and hours ago, well past the agent run time limit, and the last run time appears to be the time BEFORE it last started, (indicating that it did not complete this time, which it obviously didn’t.)
So,… Has anyone observed similar behavior? BTW - this is a Java agent, but I won’t get into what it’s doing unless someone thinks it’s relevant, since it does seem to finish everything. It does EVERYTHING except for shut down!
It sounds like the agent is finished, but JVM is having trouoble shutting down some threads after the agent is finished. So this is not a problem with the cancel command, as far as it is concerned the agent finished running.
I don’t work on JVM, but I recollect there might have been some work in 6.01 related to shutting down threads under certain condition. If you can upgrade, see how your agent performs in 6.01. Hopefully, the same situation will not occur.
This AMgr: Agent ‘agent’ in ‘db.nsf’ is not running
just happened to me too. Its a LotusScript agent that takes a long time to run. It was started with the AMgr RUN command. Re-starting AMgr shuts down the agent with the ‘completed execution’ message, though it hadn’t really finished.
