java.lang.OutOfMemoryError

[manuel-sejourne]

Hi,

We found an issue on our PRODUCTION app

java.lang.OutOfMemoryError: Could not allocate JNI Env
--
  | 1 | at java.lang.Thread.nativeCreate(Thread.java)
  | 2 | at java.lang.Thread.start(Thread.java:1063)
  | 3 | at java.util.concurrent.ThreadPoolExecutor.addWorker(ThreadPoolExecutor.java:921)
  | 4 | at java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:1328)
  | 5 | at java.util.concurrent.Executors$DelegatedExecutorService.execute(Executors.java:584)
  | 6 | at android.os.AsyncTask.executeOnExecutor(AsyncTask.java:620)
! | 7 | at com.optimizely.ab.android.datafile_handler.DatafileLoader.getDatafile(DatafileLoader.java:77)
  | 8 | at com.optimizely.ab.android.datafile_handler.DatafileService.onStartCommand(DatafileService.java:73)

Thanks in advance for solving this issue

Bests

[manuel-sejourne]

I use this version :

 implementation('com.optimizely.ab:android-sdk:1.4.0', {
        exclude group: 'com.android.support.test.espresso'
    })

Does the new version solve this issue ?

[thomaszurkan-optimizely]

1. What is your datadownload interval set to (should probably be at least a few minutes Android O has a min of 15 minutes)?
2. Since it is from a service intent, it is in a separate thread that dies so worst case, the intent should just not complete.
3. We don't have a try catch around that and we should. But, the initialize is being refactored and that call will no longer exist in its current state.

[manuel-sejourne]

Hi,

My answers:

.withDatafileDownloadInterval(60)
.withEventDispatchInterval(60)

[thomaszurkan-optimizely]

@manuel-sejourne So, it is updating every 60 seconds which is quite a lot. But, have you profiled the app to see where the memory leak is coming from? Just because it died there doesn't mean that's where the memory leak is. We have cleaned up all our resources in 1.4 and I don't think we have any memory leaks.

[thomaszurkan-optimizely]

@manuel-sejourne Question: Was it in API 24? I see some weird stuff when profiling and there are lots of stack overflow on memory leaks in this version.