Need guidance on setting correct pool size

[RahulMetangale]

Hi All,

Could you please point me to relevant documentation related to deriving pool size if available else please provide some suggestion on the same.

Environment Details:

1. Three fission environments (python)
2. Multiple models/python code that run in one of the fission environment.
3. Fission is running in pool manager mode.
4. Current pool size is 3.
5. Concurrency is set to default 500

Sometime due to lot of concurrent requests, we see some errors like "error posting to getting service for function: context deadline exceeded". We increased the pool size to 5 and now occurrence of this errors is reduced.

So i would like to now how should we arrive at ideal pool size.?

• I am assuming if pool size is 3 and if 4 concurrent requests are received for a model using that environment, 3 pods will be specialized first while fission is trying to make 3 additional pods available in pool. While this is happening 4th req will wait until specialize timeout?

Any help is greatly appreciated.