Loss of connectivity to presenter can cause viewer's browser to gobble CPU

[gabe-sky]

Open a presenter and a viewer window. Block network connectivity between the viewer and the presenter's machines. (e.g. get on an airplane.) The viewer's machine after some time goes to 100% on one CPU core. I suspect it's aggressively trying to reconnect to the presenter.

Chrome Version 53.0.2785.116 (64-bit) MacOS Sierra

[binford2k]

ooooh.

Was this a viewer window that you got by clicking the button, or was it an audience window in which you just browsed to the url and let it use follow mode? I'll test both scenarios, but I'm curious as to which one you're hitting now.

[gabe-sky]

Browsed to the URL by hand and let it use follow mode.

[binford2k]

ug, that's worse because it means that it will effect all end users; the audience. Thanks

[binford2k]

Hrm. I've had it running in this scenario for 6+ hours. Chrome got up to ~16% CPU (which is too much), but never approached 100%. Can you validate whether this is repeatable, @fnaard ?

[binford2k]

I've finally run into it locally. I'll revisit this.

[marrero984]

@binford2k Did we resolve this issue when we fixed the settimeout calls? Wasn't this being caused because the reconnect method was calling as fast as it could instead of waiting for five seconds?

https://github.com/puppetlabs/showoff/pull/784

[binford2k]

This is possible & likely. Do you know of a way to trigger it to make it more easily testable?

[marrero984]

I don't have a way of reliably triggering this