All, Is there an API to export JMX config?  I see APIs for exporting dashboards, transaction detection rules, alerts, etc, but nothing for JMX. This is where I'm looking: https://docs.appdynamics.com/appd/24.x/latest/en/extend-splunk-appdynamics/splunk-appdynamics-apis/configuration-import-and-export-api   thanks   ... View more

All, In the metric browser, I see a DB appear under Backends|Discovered backend call...  I also see the same backend under Overall Application Performance|My Tier|External Calls|Call-JDBC to Discovered backend... The calls per minute graphs are approximately the same shape, but counts are not even close.  Under backends, the counts are much higher, like 5x higher.  This app only has 1 active tier. Why is there such a large difference in counts?  I would like to get a breakdown of DB calls per tier, but the numbers seem low. thanks ... View more

The java agent doesn't always understand some thread handoffs or exit calls.  There are examples in the documentation for how to use the AppD API to handle this.  However I would like to know if it can be done in a vendor-neutral way.  For example, with other APM agents, it's sometimes possible to use OTel manual propagation and it "just works."  Can this be done with the AppD java agent? thanks ... View more

All, What does it mean when there are multiple HttpServlet:service() and dispatcherServlet:doService() nodes in the call graph of a transaction snapshot?  For example: Call graph showing multiple calls to HttpService.service. The service does use multiple threads. thanks ... View more

All, The OOTB Kafka JMX configuration is incorrect.  Maybe it was correct for an older version of the Kafka client library, but it's not correct any longer.  For example, there are metrics defined using the object name match pattern "kafka.consumer:type=consumer-fetch-manager-metrics,*" such as records-lag-max.  However records-lag-max is not an attribute of any object matching that pattern. I have fixed this problem manually and disabled the OOTB configuration.  My new configuration uses the object name match pattern "kafka.consumer:type=consumer-fetch-manager-metrics,client-id=*,topic=*" and the instance identifier "client-id,topic."  My new metrics do show up in the metric browser.  If the client id is myClientId and the topic is myTopic, the metrics show under JMX|ConsumerMetrics|myClientId|myTopic. However I cannot seem to use the new metrics in a dashboard.  When adding a time series to a dashboard, I can choose "JMX|ProducerMetrics" or "JMX|ConsumerMetrics" as the JMX objects to include.  But when selecting the actual metric to display, I can't see any of my new metrics.  I can only see the OOTB metrics like Average IO Wait.  When selecting the JMX objects to include or when selecting the metrics to display, I cannot drill down further than ProducerMetrics or ConsumerMetrics, even though there are two levels below (corresponding to the client id and topic.) thanks ... View more

All, I'm using the SaaS controller.  I'm familiar with metric rollup but there is a difference in data granularity between the SaaS UI and the output from the metric-data REST endpoint. Currently I'm looking at some data in the UI and it shows 1 minute intervals for a week ago.  However when I query the same approximate time range from the REST API, it's giving me 1 hour intervals.  Rollup is false.  I've tried adjusting the start time of my REST query to be < 1 week ago.  The time range of my query is 2 hours. Through trial and error, it looks like the API returns 1 hour granularity for times more than about 24 hours ago.  Somewhere between 12-24 hours ago, I get 10 minute granularity and 1 minute for less than that. Is there a way I can get the 1 minute granularity from REST.  Obviously the data exists if the UI is showing it to me. thanks ... View more

I'm using the v20.5 java agent. Is there a way to trigger and download a thread dump via the API?  I don't see one. In the UI, we can do something like this: Go to node Go to agent tab Request agent log files Collect x thread dumps y millis apart Download resulting files I would like to automate that. thanks ... View more

I'm using: Java HotSpot(TM) 64-Bit Server VM 11.0.9 Oracle Corporation Server Agent #20.5.0.30113 v20.5.0 GA compatible with 4.4.1.0 r474b6e3c8f55ababbb11a87ff265d8ce34eb0414 release/20.5.0 I cannot use the object instance tracker.  It gives a warning about missing tools.jar or needing to use an Oracle JVM. I see in release notes for 4.5.2 that there was a bug fixed in this area.  What other causes could there be for the object instance tracker to not work?  Does it have something to do with "compatible with 4.4.1.0?" thanks ... View more

All, Is there a way to get the java version via the REST API?  /applications/foo/nodes seems like the logical resource but doesn't have the java version.  It does have the machine and app agent type and version but I don't see the java version anywhere. thanks ... View more

I'm using agent v20.3 and CXF 3.2.12. AppD isn't identifying individual web services on the front end.  Instead, it's lumping them all in to a single business transaction named after the url like "/myapp/services."  The java auto discovery rule has "web service transaction monitoring enabled" and "discover transactions automatically for all web service requests" both checked. What else can I do differentiate the individual services? thanks ... View more

I'm using 4.5.5. I created an exit point to track calls to java.sql.Connection.isValid() but it doesn't work.  Since then I've been seeing this error in the agent logs: [Log files redacted] Any suggestions? thanks ^ Post edited by @Ryan.Paredez to remove log files. Please do not share or attach log files to community posts for security and privacy reasons. ... View more

I'm using AppD 4.5.11.3848 and Oracle JDK 1.8.0_231. I've seen some incorrect process CPU% reported by the java agent.  It appears that the number is being derived as follows: Process CPU burnt (ms/min)/60000/Runtime.availableProcessors() In other words, the number of milliseconds of CPU used in a minute, divided by the number of milliseconds in a minute, divided by the result of Runtime.availableProcessors().  So, if there are 4 CPUs and the "cpu burnt" number is 30000, then we'd have 30000/60000/4, which is 12.5% However, the Runtime.availableProcessors() method doesn't always report the actual number of processors.  Instead, if -XX:ActiveProcessorCount=n is set, it uses that.  In my example, if there were actually 8 CPUs but -XX:ActiveProcessorCount=4 is set, then AppD will report 12.5% when the actual number should be 6.25% (divide by 8 instead of 4.) The active processor count does not actually limit the number of CPUs the process can use so I think this method of calculating process CPU % fails when the active processor count is set to something other than the actual value. thanks ... View more

Hi all, I have a 3rd party library that exposes metrics via JMX.  Unfortunately some of them are simple counters: values that start at 0 and go up and up as the app runs.  Over time, the numbers get so big that looking at a graph over a short time period isn't useful because I can't see small changes in big numbers.  What I really want is the difference in the value over consecutive time periods.  That is, if the value was 1,000,000 a minute ago and is 1,000,037 now, I want 37 displayed/recorded. I don't see anything like this in the docs.  Is it possible? Thanks ... View more