Implement metrics for external queue

[Atharva-Kanherkar]

WorkStealingThreadPool Metrics Implementation

This PR implements metrics for tracking singleton and batch task submissions to the external queue in WorkStealingThreadPool, as in issue #4269 .

Changes Made:

1. Added Atomic Counters to WorkStealingThreadPool:

   • singletonsSubmittedCount - tracks total singleton task submissions
   • singletonsPresentCount - tracks current singleton tasks in the queue
   • batchesSubmittedCount - tracks total batch task submissions
   • batchesPresentCount - tracks current batch tasks in the queue

2. Updated Submission Logic:

   • Changed submitToExternalQueue() to increment counters when tasks are added
   • Changed pollTask() to decrement present counters when tasks are polled

3. Added Accessor Methods:

   • getSingletonsSubmittedCount() - returns total singleton submissions
   • getSingletonsPresentCount() - returns current singletons in queue
   • getBatchesSubmittedCount() - returns total batch submissions
   • getBatchesPresentCount() - returns current batches in queue
   • logQueueMetrics() - Helper for logging all metrics

4. Added Verification Test:
   I made a test (which i am not sure about, but It did work -->)
   

   • Submits singleton tasks and verifies the counter increases
   • Triggers batch creation through local queue overflow
   • Verifies both types of metrics track correctly

• Singleton submissions: Created 10,000 tasks → Counter increased by exactly 10,000 ✓
• Batch submissions: Generated queue overflow → Counter increased by 2,053 ✓

This might confirm that the implementation works as fine as it should be. The code too compiles without any errors, at least in my terminal :)
The test is included in the PR for reference, showing how to access these metrics from client code.

[Atharva-Kanherkar]

Hey @armanbilge! Please ignore some of the previous commits as i got myself into a git command hell :p , Anyways, the PR is now ready to review. Please only check the recent commit. It will clearly show what the code I have changed. Thanks :)

[armanbilge]

Thanks for your work on this!

May I suggest an alternative approach? I think we should make ScalQueue no longer generic. Internally, it can be backed by queues of AnyRef. Then, we can adjust its API to be:

def offer(a: Runnable, random: ThreadLocalRandom): Unit
def offer(a: Array[Runnable], random: ThreadLocalRandom): Unit

Then, we can keep the counters inside of ScalQueue, and increment the appropriate counters in the specific offer methods.

poll will have to return AnyRef and use isInstanceOf checks to decrement the appropriate counter.

[Atharva-Kanherkar]

Thanks for your work on this!

May I suggest an alternative approach? I think we should make ScalQueue no longer generic. Internally, it can be backed by queues of AnyRef. Then, we can adjust its API to be:

def offer(a: Runnable, random: ThreadLocalRandom): Unit
def offer(a: Array[Runnable], random: ThreadLocalRandom): Unit

Then, we can keep the counters inside of ScalQueue, and increment the appropriate counters in the specific offer methods.

poll will have to return AnyRef and use isInstanceOf checks to decrement the appropriate counter.

For sure. Give me a day, i will look into it!

[Atharva-Kanherkar]

Hey @armanbilge :) I have added the requested changes. Please review! If there is anything to fix again, please let me know!

[armanbilge]

Thanks for making those changes! It looks like the code doesn't compile currently.

[Atharva-Kanherkar]

Thanks for making those changes! It looks like the code doesn't compile currently.

Yes! I am sorry, I actually slept after pushing the changes and forgot fixing the issues. I will be doing it right away! ~Atharva

[Atharva-Kanherkar]

@armanbilge Actually there were so many references and inter dependencies of how scalqueue was generic, that is why this happened. It seems like there is a lot of code which is written kept in mind that it is generic. However, Ill fix it soon:)

[Atharva-Kanherkar]

Update :

I guess the code works as we wanted it to work! There are still warnings and formatting issues, which can be fixed promptly, but the functionality is working fine now.

[Atharva-Kanherkar]

Hello @armanbilge ! I have added the changes! Please review :)

[Atharva-Kanherkar]

Hey @armanbilge!
Just wanted to follow up on this PR — marked it as ready a couple days ago. No rush at all, but when you get a chance, your review would be super appreciated! 🙌

[Atharva-Kanherkar]

hey @armanbilge ! do you need anything else from my side! :) happy to help..

[armanbilge]

@djspiewak do you have any performance concerns here, should we run some benchmarks?

[djspiewak]

I would definitely like to run some benchmarks on this one. I'll review the diff more closely though.

[Atharva-Kanherkar]

Hi @armanbilge @djspiewak! Hope you're doing well. Just wanted to follow up on this—any updates by chance?