High Response Time on Heroku with One Click deploy

I just installed Rocket Chat on Heroku using the One-Click Deploy. Everything seems to work fine at first sight except I get “High Response Time” alerts from Log Entries. Example:

2018-02-01 10:37:40.191 340 <158>1 2018-02-01T10:37:39.437529+00:00 heroku router - - at=info method=GET path=“/sockjs/582/7ychqe79/websocket” host=acme-rocket-chat.herokuapp.com request_id=3aa62012-d535-4d7d-a952-7140cf304bf4 fwd=“84.123.67.6” dyno=web.1 connect=0ms service=307141ms status=101 bytes=2497 protocol=https

It looks like it’s related to websockets. Is there anything I should check

What’s your user load look like when you are getting this?

I am having this same problem, but with only 3 users online.

2018-05-24 20:28:29.097 314 <158>1 2018-05-24T20:28:28.906083+00:00 heroku router - - at=info method=GET path="/_timesync" host=***REMOVED*** request_id=e6b9c32d-6d12-4fee-8fa1-d9234207abae fwd="104.200.154.106" dyno=web.1 connect=1ms service=2ms status=200 bytes=282 protocol=https
2018-05-24 20:29:23.132 339 <158>1 2018-05-24T20:29:22.968771+00:00 heroku router - - at=info method=GET path="/sockjs/926/av2_2akm/websocket" host=***REMOVED*** request_id=acab588a-2bd9-45aa-8acd-11233e270c14 fwd="207.173.82.116" dyno=web.1 connect=1ms service=2018233ms status=101 bytes=175 protocol=https

Interesting… Wonder if the dynos don’t like the long running websocket connections. I know the free throttles CPU and such. Might be related.

How do mobile apps work with it?

This is a consistent problem we have been dealing in Heroku, and is related with sockjs. Apparently there is a workaround for this in this url https://gist.github.com/majek/2427902 but I’m not fully able to understand how to try/port it in meteor.

I would suggest to request Heroku to put an eye on this because is affecting every single instance running Meteor, even when there arent users connected.