How can I do this? Capture Timeout
Theory: Some transactions went to timeout and did not even reach the webserver back (I see most of them in DT but not able to find the slow ones in the webserver logs...). They kept tying from JVM to JVM...
Question: How can I leverage dynatrace to capture such incidents. These transactions had a big impact on our production response times and almost caused an outage.
One such sample given below.
Jibi.
Hi Jibi,
I believe the Timed Out PurePath Incident in the System Profile would be of interest to you here.
This doc might be of interest to you, PurePaths#PurePathsTab, there are some notes at the bottom of the PurePath States dealing with timeouts.
Thanks,
David
David - Thanks. Can you attach the doc again? Link seems to be not working...
For my app I changed this time out to 360 seconds
-Dcom.dynatrace.diagnostics.completenessTimeout=360
Wondering if User Action Response time can be used for any such monitoring? After what time would a JS agent time out????
The timing out of PurePaths is more a feature to avoid overloading the server if many PurePaths take a long time. Also the related incident is more a self-monitoring incident and is not very customizable when having a diverse set of PurePaths being recorded.
Usually you would use one of the available measure with according thresholds. Either something like "PurePath Response Time" for all PurePaths recorded in the system profile or e.g. a measure based on some code-method for a more fine grained configuration. There are many more types of measures that you can create, e.g. based on URI, WebService, Remoting, Messaging, JNDI, ...
Then you can define an incident on this measure which alerts you, this also allows things like alerting different people based on which type of method/URI/WebService took too long, ...
