Service > Instances pulsing forever when /mesos/master/state is large

Clicking on Services > Instances can result in the tab pulsing forever and not loading.

 

When logging into the UI and clicking Services, there will be an XHR call fetchState() to /mesos/master/state?_timestamp=xxxxxx

On a largish cluster with a slowish client connection, this XHR call can take a while.

Example from my customer's cluster with a slow client connection, it took 28 seconds

`time curl -vkL -H … https://<master>/mesos/master/state?_timestamp=1511366906328 >out 2>err`

gives result of
real    0m28.839s with stdout of size 2.1m and stderr of size 4.1k

This exceeds the timeout of 2000 milliseconds set at https://github.com/dcos/dcos-ui/blob/master/src/js/events/MesosStateActions.js#L15

As a result, the XHR call fails and the Instances tab never loads to completion.

I have increased the timeout to a more generous 200000 milliseconds and the issue went away and we could load the Instances tab just fine.