cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Application process shutdown incident in .NET

gopikrishnanr
Organizer

Hi,

There were many instances where a lot of application process shutdown incident was raised and when we try to verify this from client the application was found to be working. My agent is listening to process in .NET 4 and installed in windows server 2008. When we check the collector logs its found that agent unregisters and re-registers itself after sometime but with new pid. Have anyone faced similar issue?

10 REPLIES 10

Babar_Qayyum
Leader

Hello Gopikrishnan,

Once again the same question that do you have any firewalls in between agents and collector.

  • The AppMon Collector performs the byte-code modifications when instrumented applications start up. Therefore, the AppMon Agents need to send the byte code of your applications to the Collector. To keep start-up times low, it is necessary to deploy at least one Collector in every location/data center that is connected to an AppMon Server.
  • A firewall introduces latency in the calls between the Agent and Collector. This is often the reason for slow application start-up. The Agent needs to do several 10,000 round trips to the Collector at application start up. Even 1 ms firewall latency will add up to a noticeable time. Therefore, either use a real fast (in latency time) firewall or put the Collector into the same subnet as the Agents.

Regards,

Babar

Hi Babar,

Thanks for the response.

I have the tracert snap attached as below. There is negligible latency as seen

Also as I said there agent unregisters and registers itself with new pid, so even if there is latency the pid should remain the same I think.

Regards,

Gopikrishnan R.

Hello Gopikrishnan,

When the agent reconnect again to the collector then it will have a new PID.

What is the sizing of your collector?

What is the Buffer Saturation, CPU Usage and Memory Usage of the collector?

Regards,

Babar

Hi Babar,

I have confirmed that there is no firewall between agent and collector.

Below are the snaps for sizing :

Regards,

Gopikrishnan R.

Hello Gopikrishnan,

One more following check to suspect any
latency.

  • Open the 'Agents Overview' dashboard
  • Add the below two columns after right click on the header
    • Average Instrumentor Response Time (ms)
    • Maximum Instrumentor Response Time (ms)

Share the screenshot
result.

Regards,

Babar

Hi Babar,

Its blank.

Regards,

Gopikrishnan R.

Hello Gopikrishnan,

If you are facing the same issue then I would recommend to open a Support Ticket.

Regards,

Babar

Hi Babar,

I have already raised a ticket with id SUPDT-35658. No support from dynatrace since version 6.2.

Regards,

Gopikrishnan R.

JamesKitson
Dynatrace Leader
Dynatrace Leader

As far as I know the PID you see is for the application process itself not the agent - the agent is running as part of the application process in a sense so if you are seeing a new PID my expectation is that the application has restarted or something a long those lines (there is no 'agent' process except for host agents). Do you have access to the application logs to see if there is something like that occurring?

I also don't see any reference to the agent logs, anything in there about the restart/shutdown? The shutdown incident should* only fire when there is a clean shutdown of the agent/application process detected.

James

Hi James,

I don't have access to application logs. Currently we have disabled the incident. When this was observed what we did was we verified the same as you said in the application server logs and found that there was no shutdown or restart at the time of incident but when the incident was raised there was no entry in the application logs for that time. May be the CLR when idle stops the process and re-invokes with new process when any hits are detected, can this be the case?

Regards,

Gopikrishnan R.