Our application has a custom method for performing a fullSync that does not use the couch db replication.
We have recently found that occasionally the pouch db on the device will attempt a fullSync on a device with some initial data which involves calls to _changes endpoint with ?since=0 which takes a very long time and never succeeds.
Is there way to make the liveSync[sync with live: true] error out when checkpointer returns 0 as the starting point?