We have a deployment slot that we deploy to, which we then warm up, and then swap with the live slot so that we can achieve zero downtime deployment on PaaS. This works very well in general.
However, we are noticing an issue where both of the slots appear to be responding to the event queue and processing things like
item:saved events and updating the indexes and weird things are happening like duplicate documents ending up in the index.
Currently, it seems like the best option is to be very diligent about stopping the staging slot and then rebuilding the indexes once the staging slot is turned off.
Has anyone dealt with this issue before?