Number of slots used by worker averages to just 1 where as number of slots available is around 200

We are seeing the schedule to start latency is higher with p99s going to 4s. During our perf test runs we are seeing that the number of slots used is around 1 but the available metric shows that we have around 200 available.

Really curious how our worker is not utilizing the task slots. We have tried increasing the number of pollers and concurrent execution size but the issue remains the same