And the process crashed. The process was doing some outbound web api calls with HTTPosion prior to the crash, and I guess it hit some tricky spot. I understand web API can fail, however, why does it fail this way? and am I supposed to catch this?
The messages look like the ones in this Hackney bug:
I’ve seen these infrequently in production (one per thousands of outgoing HTTP requests) so it’s hard to troubleshoot, but AFAIK they are “leftovers” from previously-used connections and don’t affect operations. To stop them crashing my GenServers, I added a handle_info that logged about it and went on with life.