gusty current
I have 4 session hosts, currently running agent 1.0.7033.1401, standard_B12MS. The Windows OS is kept up to date on a bi-weekly basis by Automox, and to my understanding the agent can only be updated automatically (there is a new version from July).
With no apparent pattern I keep seeing the individual sessions hosts go to disconnected. This coincides with LogicMonitor reporting that the CPU on the machine in question has maxed out.
Once this has happened it also becomes impossible to log in to the machine. I redeploy the machine, it works normally again afterwards, and then anywhere between a few hours to a few days later it will do it again.
There are also larger periods of time where the desktops will run butter smooth for a month or a month and a half, and then it goes back to doing this for another month or more.
Event Viewer on the hosts doesn't seem to give any smoking guns. I see a huge number of services getting Timeout errors around the crash, preceded by the AzureDiagonistics noting the Heart Beat write failed. But there's nothing to indicate why that happens. The Virtual Disk Service is constantly starting and stopping, but I'm not sure if that's normal behavior.
Has anyone seen something like this or can give further advice on how to tease out more definitive information on the nature of the problem?
I was thinking of trying to capture more information with the Data Collection Rules, but I've never done that before. Not sure if it'll give me what I want--to definitively confirm which process is maxing out the CPU.
hoary depot
austere mantle