Silos unable to update membership table while pod is terminating

[08Leinad08]

The behaviour has changed in the latest version of Calico AKS has adopted, whereby terminating pods have their network connectivity removed immediately. As a result of this, we've found that silos are unable to update the membership table (in Azure) before shutting down gracefully, causing them still to be active in the membership table.

We haven't found this to be too much of an issue when scaling down to 3 or more silos. Assuming the remaining active silos work together in order to correct the table. However for scaling down to 1 or 0 silos, this takes a lot longer for the silo to self-correct the membership table and can crash.

Azure/AKS#2223
projectcalico/calico#4518

This is more of an AKS issue, but wondered if others have come across this?

Thanks