Non-unique Agent requested connection

[pivst01]

hi all,

 

we have server restarts at each night and (i think) because it's up again in very short time, the agent cant connect under the same agentname as it was connected before. so new "agentnames" are created and so a lot of historical metrics.

 

5/20/15 01:20:09.376 AM CEST [INFO] [PO:WatchedAgentPO Mailman 5] [Manager.Agent] Connected to Agent "SuperDomain|CZZSP0W103|.NET Process|core" at "Node=Agent_284, Address=s08984mw.cz.srv.sys/10.215.16.13:49434, Type=socket" using Agent protocol revision 1.0

5/20/15 01:20:10.711 AM CEST [WARN] [PO:WatchedAgentPO Mailman 4] [Manager] Non-unique Agent requested connection ("SuperDomain|CZZSP0W103|.NET Process|core"). Re-assigning Agent Name: core%2

5/20/15 01:20:10.750 AM CEST [INFO] [PO:WatchedAgentPO Mailman 4] [Manager.Agent] Connected to Agent "SuperDomain|CZZSP0W103|.NET Process|core%2" at "Node=Agent_285, Address=s08984mw.cz.srv.sys/10.215.16.13:49435, Type=socket" using Agent protocol revision 1.0

5/20/15 01:20:23.033 AM CEST [INFO] [PO Async Executor] [Manager.Agent] Disconnected from Agent "SuperDomain|CZZSP0W103|.NET Process|core"

5/20/15 01:20:30.808 AM CEST [INFO] [PO Async Executor] [Manager.Agent] Disconnected from Agent "SuperDomain|CZZSP0W103|.NET Process|core%1"

5/20/15 02:20:27.060 AM CEST [INFO] [TimerBean] [Manager.Agent] Automatically unmounting Agent "SuperDomain|CZZSP0W103|.NET Process|core" after no communication for 60 minute(s).

5/20/15 02:20:27.061 AM CEST [INFO] [TimerBean] [Manager.Agent] Unmount completed of Agent "SuperDomain|CZZSP0W103|.NET Process|core"

5/20/15 02:21:27.060 AM CEST [INFO] [TimerBean] [Manager.Agent] Automatically unmounting Agent "SuperDomain|CZZSP0W103|.NET Process|core%1" after no communication for 60 minute(s).

5/20/15 02:21:27.061 AM CEST [INFO] [TimerBean] [Manager.Agent] Unmount completed of Agent "SuperDomain|CZZSP0W103|.NET Process|core%1"

 

do you have any tip/solution how to configure the agent/EM

- to either release the agent (and it's name) in shorter time after the agent disconnection

- or wait for longer time and allow the same agent to connect to EM under the same agentname as it was connected earlier.

or any other tip/solution how to fix this behavior.

 

the agent naming process should be automated as there are hundreds of agents and so to name them manually on each server(profile) is not possible.

 

thank you,

stefan

[Lynn Williams]

Hi Stefan,

1. The EM will automatically unmount the agent but only after the period set by this EM property which defaults to 60 minutes: introscope.enterprisemanager.autoUnmountDelayInMinutes, so you could try reducing that to a few minutes. However the downside is that if you do get a transient agent -EM connection problem while the agent is up then there is more chance that it will be unmounted, which is why I think the default is 60 mins. Also if you do reduce it then introscope.enterprisemanager.agentconnection.metrics.agentTimeoutInMinutes should also be <= that value
2. Alternatively from the investigator you can manually unmount the agent after it has stopped (there is also a CLW Field Pack extension available for script based equivalent).
3. Depending on your APM version there is also a known problem with % suffixed names still appearing for 9.5.x (fixed in 9.6 & 9.7). See this previous discussion for details:Automatic unmounting of agent fails

 

Hope this helps!

 

Regards,

 

Lynn

P.S.

This KB article may also be useful:

APM Investigator displays Agent Name with percent sign (%) and a number.