Android ANR troubleshooting with MINT

Updated 2/19/2021: Splunk MINT is no longer available for purchase as of January 29, 2021. Customers who have already been paying to ingest and process MINT data in Splunk Enterprise will continue to receive support until End of Life, announced as December 31, 2021 for all MINT products: App, Web Service (Management Console), SDK and Add-On.

Being involved with shippable software for mobile and desktop, I realize that there is a class of problems that are not easy to troubleshoot.

Crashes are probably the easiest to reproduce in QA and Engineering environments and so they are easier to fix. But one class of problems, that in many cases requires more time and possible code redesign, is application sluggishness. This problem usually falls into the gray area of software development that everybody tries to address during design and implementation stages. The problem of application sluggishness seldom shows up in QA or other controller environments, but always happens when the actual user is trying to use the app.

Modern mobile apps are complex creatures. A lot of things are happening as a result of user input or internal processes in the background that are also trying to update the UI. Apps can also issue many backend calls to keep the UI up to date. 

We all like a smooth UI experience with our apps. Android addresses UI issues by implementing an Application Not Responding (ANR) mechanism, which forcefully terminates non-responding apps. The timeout is enforced by the system and the data is available in the LogCat.

In the 5.1 release of the Splunk MINT SDK for Android, we’ve given you a way to monitor and troubleshoot your app’s ANR issues. Just opt-in for ANR monitoring for your app by calling:

Mint.startANRMonitoring(5000/*timeout*/, true/*ignoreDebugger*/);

ANR events will then be available in Splunk Enterprise. Run this search to view them:

sourcetype="mint:error" "extraData.ANR"=true


Splunk Enterprise ANR event samplePlease note that the stacktrace field in the event should be interpreted as a thread dump of your application threads (see the link to the documentation and example below). 

Our monitoring feature will help you to identify common problem of ANR, such as application deadlocks and unexpectedly long-running or stalled HTTP requests.

Additional Reading: 


Michael Margulis

Posted by