Don't exit travis monitor if a server error happens

[sarahhodne]

Right now, if Pusher or our API throws a 500 error, travis monitor will just exit. In the case of a Pusher error, we should probably reconnect to the stream, and maybe retry (or silently ignore) if our API errors?

[rkh]

👍

related: pusher-community/pusher-websocket-ruby#23

[Habbie]

I found my "travis monitor" eating all of one CPU core. Last output:

2013-11-05 14:51:19 PowerDNS/pdns#1048.2 passed
2013-11-05 14:51:28 PowerDNS/pdns#1048 passed
2013-11-05 14:51:28 PowerDNS/pdns#1048.1 passed
F, [2013-11-05T16:22:50.369657 #28937] FATAL -- : Pusher : error : {"code"=>4200, "message"=>"Please reconnect immediately"}

so +1 on pusher-community/pusher-websocket-ruby#23 affecting travis I guess.

[rkh]

It did? Hmm… I've never seen that happen. Curious as to why it would go crazy on CPU if the pusher connection is lost.