Domino on Linux/Unix, Troubleshooting, Best Practices, Tips and more ...

Troubleshooting Agent Manager Abnormal Process Terminations

Daniel Nashed  17 March 2021 14:27:13

During a Domino 11.0.1 migration on AIX we ran into an abnormal process termination.

This is the most ugly situation you can run into, because the process will just crash and often not leave any hint behind what went wrong.
Depending on the platform you either see the parent process cashing with a "child died" signal (Linux/UNIX) -- in this case the agent scheduler process -- or the server process monitor crashing (W64).

In some cases you can still find information in memcheck about the running agent.
Finding the info there is quite complicated.

There is an easier way to find the current running agent.
Before the agent manager starts an agent, the agent database and name is written into the server statistics.
The server statistics are in memory and are dumped by NSD.

In our case it turned out that a Java agent was running.

You will see the following type of information in the server statistics -- included in NSD.

Example output:

Agent.C80142.AgentDBPath = my-application/my-database.nsf
Agent.C80142.AgentName = MyCrashingAgent

Each Amgr executive will write a separate entry into your server statistics.
After the first crash we set the number of Amgr executives to one for better tracing and found the crashing Java agent.

I hope you will never run into this situation.
But at least now you know what to search for.

-- Daniel



Links

    Archives


    • [IBM Lotus Domino]
    • [Domino on Linux]
    • [Nash!Com]
    • [Daniel Nashed]