CF V3 getProcessStatistics - The threads are not released

[benuantony]

@twoseat I am tagging you directly as the issue still exists.
we have about 70 microservices running in the cloud foundry & we would like to monitor all the microservices with the following metrics,

• State
• Memory utilization
• CPU Utilization
• Disk Utilization

recently, I started to use CF v3 API & the issue is still the same. The allocated threads are not released, Due to which the application crashes multiple times a day.

The below API is called once every 5 minutes for about 70+ microservices

    public void requestAllApplicationStatistics(DefaultCloudFoundryOperations cloudFoundryOperations, List<CfArtifacts> cfArtifacts) {
        initializeTaskExecutor();

        Flux.fromIterable(cfArtifacts)
                .flatMap(application -> cloudFoundryOperations.getCloudFoundryClient().applicationsV3()
                        .getProcessStatistics(GetApplicationProcessStatisticsRequest.builder()
                                .applicationId(application.getId())
                                .type("web")
                                .build())
                        .map(response -> {
                            logger.debug("step 1 - cfStatistics fetched successfully");
                            CfApplicationResponse cfApplicationResponse = new CfApplicationResponse();
                            cfApplicationResponse.setApplicationId(application.getName());
                            cfApplicationResponse.setCfApplicationId(application.getId());
                            cfApplicationResponse.setPrefix(application.getPrefix());
                            List<ProcessStatisticsResource> resourceList = response.getResources();
                            cfApplicationResponse.setApplicationInstances(resourceList);
                            logger.debug("step 2 - cfStatistics before calling the callback");
                            return cfApplicationResponse;
                        })
                        .doOnError(e -> {
                            logger.error("cfStatistics Fetch Failed with error: {}", e.getMessage(), e);
                        })
                        .doFinally(signalType -> {
                            logger.debug("All applications processed. Shutting down the executor service.");
//                            shutdownTaskExecutor(); // Shutdown the executor service
                        })
                )
                .subscribeOn(Schedulers.fromExecutor(taskExecutor))// Use the shared task executor
                .publishOn(Schedulers.fromExecutor(taskExecutor)) // Use the shared task executor
                .doOnNext(cfResponseCallback::processCfMetrics)
                .subscribe(); // Subscribe to start the execution
    }

    private static synchronized void initializeTaskExecutor() {
        if (taskExecutor == null) {
            taskExecutor = new ThreadPoolTaskExecutor();
            taskExecutor.setCorePoolSize(10); // Set the desired core pool size
            taskExecutor.setMaxPoolSize(20); // Set the desired maximum pool size
            taskExecutor.setThreadNamePrefix("cf-task-executor-"); // Set a thread name prefix
            taskExecutor.initialize(); // Initialize the task executor
        }
    }

Look at the active threads in the below screenshot, it keeps increasing & crashes (the screenshot was captured from the Health Monitoring on the total threads measured by java Open telemetry instrumentation)