Reconnect to a Socket UDF after crash/error?

We are running into an issue where Kapacitor does not retry connecting back to a Socket UDF when there is a crash or an error is returned.

If the Socket UDF panics and the connection to the Unix Socket file is severed, Kapacitor will close the socket server and never reattempt connection and stops the task execution altogether.

Does Kapacitor have a better way of handling this problem? Is it possible we missed a configuration setting that allows exponential retry if a connection is severed? This is currently a big problem for us as we are considering Kapacitor to be our central alarming system but with this behavior, we cannot go forward with this tool. We have a possible workaround which would be to track and restart all failing tasks however this would cause downtime and possible loss of alerts.

Start by tracing the Socket UDF connection lifecycle and the behavior after the Unix socket is severed by a panic or returned error. Reproduce the failure and determine whether reconnect or exponential retry is supported; done means the task resumes without manual task restarts or alert downtime.

Written by the indexing model from the issue text.