buildcache | advanced compiler accelerator that caches and reuses

After spent lots of time on this problem, I decided to upgrade package.json, at the first step I've upgrade 3 packages including "@vue/composition-api": "^1.0.0-beta.3" -> "^1.0.0-beta.10" and "sass-loader": "^7.1.0" -> "^9.0.3" and "typescript": "~3.9.3", -> "~3.9.7", and suddenly the typechecking time decreases to 4s, which was very good promising.

So, here is what i did :

1.use "skipLibCheck": true flag in tsconfig.json , which skips lib type checks which will make type-checking 2x faster.

1. if the problem still exists -> upgrade your package.json

you can use yarn upgrade-interactive --latest but be careful about breaking changes(read the docs to fix you problems, in case of sass-loader upgrading if you have issue read https://stackoverflow.com/a/62844942/12666332)

When Gradle parses gradle.properties file, too much has already happened. So the only way to use a custom Gradle home is through a command line flag, which has the highest precedence, or through a system property.

Note that the documentation indicates that and does not mention the ability to set that value inside a properties file.

The problem is that native library does not define a method that matches the signature of the Java native method; i.e. method name or signature in the JNI code is wrong ... or maybe it isn't a JNI native library at all.

As one can see the issue is from libjvm.so JVM environment and not an application problem

No: the issue is most likely with libbCacheService.so, and not libjvm.so. I don't know what the origin of libbCacheService.so is, but given that google only finds this question when I search for it, I suspect that it's your application.

Besides, the same setup and code on Ubuntu works fine

That is quite often the case: an application exhibiting undefined behavior works "fine" on one system, and crashes on another.

Your first step should be building libbCacheService.so with debug info, enabling core dumps, and checking from the core that whatever libjvm.so function is invoked from JNIEnv_::GetObjectClass() is called with correct arguments.

P.S. It looks (from si_addr) that libjvm.so dereferenced a NULL pointer. Possibly because you passed a NULL pointer to it as one of the parameters.

There is no built-in command for that, but you can make your own. I have the same requirement for one of my projects and solved it like this:

The first thing that jumps out at me is that you're not setting -XX:MaxDirectMemorySize to anything specific. This means it defaults to the same as the max heap size. Given that you're trying to create 4(?) caches with 200MB of offheap each you may well just be running out of offheap space. Try adding a suitable value when you spawn the JVM?

The gradle-enterprise-maven-extension, as it name implies, is an enterprise only extension. All of its features require a valid license, checked for by contacting a Gradle Enterprise server.

Unlike the Gradle build tool which offers its caching abilities as part of the open source solution, both local and remote, the Maven extension has no free features.

Ben Manes suggested in his comment that I use Runnable::run as the LoadingCache's executor when running unit tests, which did the trick!

I implemented a second protected buildCache method on my factory that additionally takes an Executor parameter, which my test class uses to pass Runnable::run.

The updated ErrorHandlingLoadingCacheFactory:

Usually the first thing to do when you need help with JGroups/Infinispan is setting trace-level logging.

The problem with KUBE_PING might be that the pod does not run under proper serviceaccount, and therefore it does not have the authorization token to access Kubernetes Master API. That's why currently preferred way is using DNS_PING, and registering a headless service. See this example.