velero-notifications intermittently fails

[jkanczler]

From time to time, the velero-notifications pod fails and after it just hangs, it stops producing logs.

The last log this time:

INFO -- : Reconnecting to API...
/usr/local/bundle/gems/excon-0.99.0/lib/excon/socket.rb:79:in `block in readline': EOFError (EOFError) (Excon::Error::Socket)
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/socket.rb:70:in `loop'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/socket.rb:70:in `readline'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/response.rb:73:in `block in parse'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/response.rb:72:in `loop'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/response.rb:72:in `parse'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/middlewares/response_parser.rb:7:in `response_call'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/middlewares/redirect_follower.rb:82:in `response_call'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/connection.rb:459:in `response'
        from /usr/local/bundle/gems/excon-0.99.0/lib/excon/connection.rb:290:in `request'
        from /usr/local/bundle/gems/k8s-ruby-0.14.0/lib/k8s/transport.rb:294:in `request'
        from /usr/local/bundle/gems/k8s-ruby-0.14.0/lib/k8s/resource_client.rb:210:in `meta_list'
        from /app/lib/controller.rb:64:in `last_resource_version'
        from /app/lib/controller.rb:92:in `rescue in watch_backups'
        from /app/lib/controller.rb:74:in `watch_backups'
        from /app/lib/controller.rb:43:in `start'
        from app.rb:8:in `<main>'