While the “tell amgr run” and “tell amgr cancel” are two of the better additions in Release 6, “tel amgr run” doesn’t show that the agent started this way is running.
Normally, when a scheduled agent runs, it’s assigned to one of the amgr executives that are running, and a simple “show status” will show that agent running.
Not being able to see this when using “tel amgr run” makes it quite difficult when trying to debug problems between R5 and Release 6 agents.
Hope that there will be some way in a future build for us to check the status of agents started in this fashion.
Any updates on this? Since version 7 still has no enhancements to this, what about version 8? It would be nice to be able to see the status of agents invoked by “tell amgr run…”
Thanks Julie, Of course the “run” command displays when it starts )and stops as well).
However, to monitor whether or not an agent is running, particularly if it has a lengthy run time to start with, we need the ability that “show tasks” provides, showing which agents are currently running. On a highly active server, it’s a bit overwhelming to try and watch the server console for a scrolling message that says "AMGR: Agent ‘xxxx’ completed execution.
This is possible with remote debugger… since it shows all agents no matter how they were invoked. Not a great solution, I know… Just the only thing that can do it at the moment.
Another not so pretty alternative, have your agent print out some sort of “status” so you know it is still running.
The reason why they don’t show up as show tasks, because those console agents are run in threads not as separate processes. There is no queue keeping track of them.
So when do you use these agents? What type of functionality do they do? Do you start a lot of them at the same time? I thought they would be used more like a one time utility for someone at the server console, so knowing which ones are running was not a major concern. I would be interested in learning in how people use it.
So far, I use the ‘tell amgr run’ for troubleshooting and iterative development of a scheduled agent that is running on the server. It saves me lots of time that otherwise would be wasted on setting agent schedules to 1-5 minutes, then repeatedly checking the AMgr schedule to see if it is pending or running. When I start it up myself I know its running so I don’t have to mess around with the previous method. We also would probably only ever use this on one agent at a time. However, if two were invoked at the same time and a performance problem revealed itself, how could another administrator know that the other agents were even running (without reviewing the log)?
It’s also helpful for utilties that are based on agents but lack profiles or documents for scheduling (Power Tools comes to mind; you can’t set an agent time in the application itself, only ‘enabled’ or ‘disabled’). Some folks will say you can program around this limitation, but for those of us who aren’t spending our day programming, this single command solution rocks!
Regardless, it is a VERY helpful option, and I love it!