Orkes queues not perfoming well

[sks1995]

Describe the bug
We are trying to use orkes queues in netflix conductor
As suggested in doc, changed conductor.queue.type=redis_standalone

On comparing it with dyno queues, orkes queues seemed to be under performing.
I am trying to execute 3 tasks sequentially.
Overall time taken by orkes queues is around 3s mostly, while for dyno queues, it is around 1.2s
There is huge delay b/w task being scheduled v/s task moved to in progress.
The worker in running at same frequency in both the cases.

Am I missing any configuration ?
Can both dyno queues and orkes queues dependencies be included together in classpath ?

Expected behavior
Expected orkes queues to be equally/more performant.

Device/browser

• Redis 7

Additional context
Add any other context about the problem here.

[sks1995]

@v1r3n Can you please assist ? We are in middle of load testing.

[v1r3n]

Hi @sks1995 dyno-queues and orkes-queues can be in the same path - should be fine.

I would like to know more about your setup. Can we connect and go over the setup?

[pmchung]

I think there are artificial delays added such as, most notably the second snippet:

orkes-queues/orkes-conductor-queues/src/main/java/io/orkes/conductor/queue/dao/BaseRedisQueueDAO.java

Lines 103 to 107 in 65472b5

	public final List<String> pop(String queueName, int count, int timeout) {
	// Keep the timeout to a minimum of 100ms
	if (timeout < 100) {
	timeout = 100;
	}

orkes-queues/orkes-redis-queues/src/main/java/io/orkes/conductor/mq/redis/QueueMonitor.java

Lines 159 to 166 in 65472b5

	private long getQueuedMessagesLen() {
	long now = clock.millis();
	if (now > nextUpdate) {
	size = queueSize();
	nextUpdate = now + 1000; // Cache for 1000 ms
	}
	return size;
	}