The first thing to check is the Application Log in Windows Event Viewer. Are there an abnormal number of errors coming from Cluster Controller? Also, how many cores does the machine have?
One thing you can try is resetting coordination. If that does not help, it would definitely be worth opening a case with our Support team -- including a full set of Tableau Server logs.
I'm replying so I pick up the thread... I had this exact issue a few weeks ago - after a call from support (who were/are excellent) it appears we didn't have enough cores - swapping from 2-4 has made a positive difference.
We were then able to swap to 64bit and use the 8GB of RAM, previously it was a 32 bit installation.
Others are also have the issue -
I also note that the minimum requirements for Server have changed Re: Minimum Requirements... even if others say different
I've been experiencing the same behaviour (runaway CPU for searchserver and zookeeper) on a test instance of Server 9.0.x and now 9.1.1. I've realised that in my case the trigger for this is when I stop and then restart the VM with Tableau Server on it - by saving the state of the VM, with Tableau running. (I'm actually not sure if that is supported as far as Tableau is concerned - I just do it because its a very convenient, quick way of stopping and starting Tableau.)
I've never had it happen if I shut down Tableau Server before saving the VM and restart Tableau afterwards -only if I save and resume with Tableau running.
Once it does get into that state restarting Tableau clears it. I've also found that simply killing the searchserver process and letting Tableau restart it resolves the issue. Searchserver restarts and immediately goes quiet and zookeeper also then goes quiet.
No idea if this is related to the issue others have been seeing, but thought I'd mention it in case it throws any more light on the behaviour.