Slow eventing functions deployed first with "everything" feed-boundary may starve subsequent functions on the same source bucket due to DCP backing up

An Eventing function occasionally can get into a zombied state where it will not process mutations even if undeploy’s and re-deployed.

I have seen this issue three (3) times in total, intermittently when running Eventing Handlers and doing “early terminations” while processing a large backlog (e.g. when deployed and the backlog not fully processed) a single Eventing function can get a zombie state for a particular that Eventing handler.

Where the function deploys BUT the Server/Statistics graphs for the deployed Eventing function are all zero and the handler never processes anything.

When this happens I can undeploy the zombied handler and deploy it again but the same situation occurs the stats are all ZERO and it makes no progress and processes no mutations.

But I can deploy other Functions even though the one is “stuck” or zombied the issue DOES not impact other Eventing functions…

However stopping the “couchbase-server” via “couchbase-server -k”  and restarting it will clear the issue and everything will work as expected.