Part11 discarded ActivityWorkItem 11A19122 because partition was terminated about durabletask-netherite HOT 2 OPEN

Worth noting: the orchestration is a single instance (re-uses the same instance id) and is eternal. It constantly re-schedules itself after a specific wait time.

Bringing this up because if there's some kind of memory leak in the orchestration instance somehow this might explain why we can download these 100MB files for a while and then at certain point it dies.

from durabletask-netherite.

First, let me clarify that this warning:

Part11 discarded ActivityWorkItem 11A19122 because partition was terminated
does not indicate a problem. It just means that the activity was interrupted due to a partition shutdown. In such cases, the activity will just restart automatically once the partition has restarted.

but then it seems the orchestration execution dies immediately after and restarts, then it goes into this cycle for a bit before the orchestration ultimately dies.

What exactly are you seeing here? How does the orchestration die? Any error messages or status information?

It constantly re-schedules itself after a specific wait time.

Are you using continue-as-new? Otherwise, a constant rescheduling may cause issues due to large orchestration histories.

from durabletask-netherite.