Collect the following data while the Agent Handler (AH) is in a 'max connections' state:
- Perfmon log - It must include the AH transition from a 'non-max connection' state to a 'max connections' state.
Related article: KB77680 - How to collect recorded performance monitor data with ePolicy Orchestrator counters.
- Server.log - Collect from the AH for the same time period as the perfmon log. It must include the AH transition to a 'max connections' state.
NOTE: You might need to adjust the log size to collect this data. For instructions, see KB56207 - How to enable Log Level 8 for ePolicy Orchestrator troubleshooting.
- A process dump of the Apache process while the AH is in a 'max connections' state:
- Open the Task Manager on the ePO server or AH computer that is in the 'max connections' state.
- Find the two Apache processes on the Details tab.
- Right-click the Apache process that is using more memory, and select the option Create Dump File. A (.dmp) file is created, and a message appears indicating where this file is located.
- Minimum Escalation Requirement (MER) tool output from an AH experiencing the problem.
- If you restart the AH services, does the issue return immediately or does it take time? If it takes time, about how long does it take?
- How many agents does the ePO server manage? How many are connecting to the specific AH with the max connections issue?
- What is the value configured for the agent-to-server communication interval (ASCI)?
Justification for data requested:
- MER from the server is needed in case the logging alone doesn’t fully explain the issue. Many times, other data in the MER is needed, including but not limited to:
- Environmental variables
- Running processes
- Other managed products installed
- Exact version numbers for the agent, ePO, or DLLs
- Perfmon log.- Provides key data, including the number of completed ASCI requests per second. This data provides insight into the nature of the problem.
- Process dump - The Apache process shows the threads that are in use by this process. In the case where these threads are hung, we can:
- See what these threads are trying to do
And
- Possibly why they’re hung unresponsive state
Example of the type of error recorded in the Server_servername.log file:
NOTE: The log file is stored in the
ePO_installation\db\logs folder.
E #06504 MOD_EPO mod_epo.cpp(330): Server is too busy (245 connections) to process request