com.tns.Runtime.runWorker exception

[shiv19]

Fatal Exception: com.tns.NativeScriptException: Cannot find runtime for id:1
       at com.tns.Runtime.runWorker(Runtime.java)
       at com.tns.Runtime.access$1500(Runtime.java:39)
       at com.tns.Runtime$WorkerThread$1.run(Runtime.java:472)
       at android.os.Handler.handleCallback(Handler.java:873)
       at android.os.Handler.dispatchMessage(Handler.java:99)
       at android.os.Looper.loop(Looper.java:214)
       at android.os.HandlerThread.run(HandlerThread.java:65)

nativescript: 6.2.2
tns-core-modules: 6.2.1
tns-android: 6.2.0
tns-ios: 6.2.0

I get this crash occasionally while developing with HMR.
Now, this crash is even showing up in production.
It's related to worker threads not loading as expected.

[bradmartin]

I have these showing in production here https://sentry.io/share/issue/ec9dbb43caf44eff917193994efa842a/ with no idea what is causing it at the moment.

[darind]

Could you show the worker implementation causing this issue and the way this worker is created? Are you calling worker.terminate() somewhere in your code?

[shiv19]

@darind
We're not using any custom workers in the app.
But there is only 1 plugin in our app that uses workers, nativescript-particle plugin. It has 2 workers.

I'm attaching the custom pack of the particle plugin that we use.

nativescript-particle-2.2.4.tgz.zip

[bradmartin]

We have kinvey-nativescript-sdk dep in our project which included a couple workers... I believe that's the only workers in our project, which we don't explicitly call.

Project package.json here and I don't recall any of these having workers, so it might only be Kinvey.

https://github.com/Kinvey/nativescript-sdk/blob/master/kinvey-nativescript-sdk.android.js

[shiv19]

This is the crash I get during development

An uncaught Exception occurred on "W1: ./ffa39720cb93dc65b0d3.worker.js" thread.
Cannot find runtime for id:2

StackTrace:
	at com.tns.Runtime.runWorker(Native Method)
	at com.tns.Runtime.access$1500(Runtime.java:39)
	at com.tns.Runtime$WorkerThread$1.run(Runtime.java:472)
	at android.os.Handler.handleCallback(Handler.java:873)
	at android.os.Handler.dispatchMessage(Handler.java:99)
	at android.os.Looper.loop(Looper.java:193)
	at android.os.HandlerThread.run(HandlerThread.java:65)

and on the console

System.err: An uncaught Exception occurred on "W1: ./ffa39720cb93dc65b0d3.worker.js" thread.
System.err: Cannot find runtime for id:2
System.err: 
System.err: StackTrace:
System.err:     at com.tns.Runtime.runWorker(Native Method)
System.err:     at com.tns.Runtime.access$1500(Runtime.java:39)
System.err:     at com.tns.Runtime$WorkerThread$1.run(Runtime.java:472)
System.err:     at android.os.Handler.handleCallback(Handler.java:873)
System.err:     at android.os.Handler.dispatchMessage(Handler.java:99)
System.err:     at android.os.Looper.loop(Looper.java:193)
System.err:     at android.os.HandlerThread.run(HandlerThread.java:65)
System.err: com.tns.NativeScriptException: Cannot find runtime for id:2
System.err:     at com.tns.Runtime.passExceptionToJsNative(Native Method)
System.err:     at com.tns.Runtime.passUncaughtExceptionToJs(Runtime.java:82)
System.err:     at com.tns.NativeScriptUncaughtExceptionHandler.uncaughtException(NativeScriptUncaughtExceptionHandler.java:34)
System.err:     at java.lang.ThreadGroup.uncaughtException(ThreadGroup.java:1068)
System.err:     at java.lang.ThreadGroup.uncaughtException(ThreadGroup.java:1063)
System.err:     at java.lang.Thread.dispatchUncaughtException(Thread.java:1955)

Also attaching logcat log
log.txt

[shiv19]

I started getting this ever since I started using {N} 6.2

[DickSmith]

I've been using Workers probably longer and more often than most, and never seen this in Prod. However, ever since 6.2 enabled HMR by default, just for good measure I always pass the --no-hmr flag on release builds, and also block it in the webpack.config.js:

  const {
    // The 'appPath' and 'appResourcesPath' values are fetched from
    // the nsconfig.json configuration file.
    appPath = "src",
    appResourcesPath = "App_Resources",

    // You can provide the following flags when running 'tns run android|ios'
    aot, // --env.aot
    snapshot, // --env.snapshot,
    production, // --env.production
    uglify, // --env.uglify
    report, // --env.report
    sourceMap, // --env.sourceMap
    hiddenSourceMap, // --env.hiddenSourceMap
    // hmr, // --env.hmr,
    unitTesting, // --env.unitTesting
    verbose, // --env.verbose
    ci, // --env.ci
    snapshotInDocker, // --env.snapshotInDocker
    skipSnapshotTools, // --env.skipSnapshotTools
    compileSnapshot, // --env.compileSnapshot
  } = env;

  const hmr = false;

So seems like HMR is begin enabled by default without that flag, as I have both a permanent Worker (Analytics) and temporary Worker (background downloads). And I've never had this in Prod for either, despite having ~550K active installs right now and many other crashes of other types.

[bradmartin]

I'm gonna enable the --no-hmr in our release scripts now to see if that helps with prod release builds 👍

[shiv19]

I've made a release with --no-hmr, will report here if I see that crash go away

[shiv19]

@darind
There is only one plugin in our app that uses workers.
Nativescript-particle. It has 2 workers. Attaching the latest version that we use.
nativescript-particle-2.2.4.tgz.zip

It looks like the app reaches a point where it is trying to read from a file that does not exist.

[shiv19]

Sorry, I forgot to leave an update here.
The issue still persists in production. Building with --no-hmr did not help.

[shiv19]

I tried going back to an older version of dev-webpack (1.2.1), that did not help. The crash still occurs.

[shiv19]

Every time this crash occurs, it is referring to a file that does not exist in the platforms folder.

[shiv19]

Once it starts happening, the only way to stop it is to reinstall the app. This has been frustrating to work with.

@darind is anyone being assigned to this issue?

[darind]

@shiv19, we are still unable to reproduce it locally. Can you share a project exhibiting this behavior?

[shiv19]

@darind

Can you reach me over email please, I'll be happy to share code privately.
sp@shiv19.com

[darind]

Can you try with the nativescript@next and nativescript-dev-webpack@next packages? There have been some fixes in those modules (scheduled for the 6.3 release) which should improve workers handling in HMR scenarios.

[shiv19]

@darind
I can confirm that this crash does not occur in my project while using nativescript@next and nativescript-dev-webpack@next, moreover now the HMR is working even when there is worker in the project. Very happy with this.
Closing this issue now, thank you <3

[shiv19]

Re-opened because the runWorker crash does eventually occur even in the @next build.

[shiv19]

@darind,
Have sent you an email with instructions to re-create this crash, hope it helps.

[darind]

Try disabling codeCache in your app/package.json: #1554

[shiv19]

@darind
Turning off codeCache in app/package.json resolved the issue :)
Thanks a lot

[shiv19]

@darind
This crash still exists in {N} 6.3
https://github.com/NativeScript/android-runtime/blob/05a2a3d6fa62fea7f2b6bc4208cd61f50ec4a4e2/test-app/runtime/src/main/cpp/Runtime.cpp#L239

This method even has a TODO on it

// TODO: Pete: Why do I crash here with a JNI error (accessing bad jni)