Workers liveness monitoring

Hi,
I’m trying to publish a metric indicating workers liveness through the DescribeTaskQueue API. The idea is to see if specific queue has any active pollers.
I see that a poller can still exists even though the worker is down, but with old LastAccessTime. I also noticed that when a worker is occupied with a long running task it does not update the poller LastAccessTime.
So my question is, what is the exact behaviour of the poller LastAccessTime?
And perhaps there is a better way to determine a worker liveness?
thanks!

The poller polls only when it has a slot available to process a received task. If the worker is full, it doesn’t poll, so the LastAccessTime is not updated.

Understood, thanks.
I will pursue another approach, probably either pblishing a custom metric or blackbox probe on the SDK prometheus endpoint.