Lets say I accidentally open a dataflow (power query online editor), and it spawns a large workload of mashups to start running on the on-premise gateway. These workloads are spawned in the form of "gateway containers".
Lets assume it is a production environment and we want to urgently cancel the mashup containers before they cause excessive problems for our shared resource managers (like databases and web servers).
Consider that there may be SOME "legitimate" mashup containers running on the same gateway server, and I'm trying not to harm any "legitimate" mashups. I only want to kill the mashups that I spawned by accident. And these mashups won't seem to die when I close the offending web browser.
How would I go about killing ONLY the "accidental" mashups?
My first impression is that I can look for all the process ID's of the mashups that are currently running, and cross reference that against the "CreateContainer" entries in the logs with an identity of PowerQueryOnlineCachingPool.
eg."Action":"ContainerPoolContainerFactory/CreateContainer","identity":"PowerQueryOnlineCachingPool"
(These are different than scheduled mashups which have an identity of null.)
QUESTION - Once I've identified the accidental mashups, can I simply start killing these in the task manager? Is there a long-term consequence? Will they be auto-restarted?
... Sorry if this sounds like a brute-force technique, but I don't think there is any other way to reliably solve the problem. I've found that it can even be difficult to avoid these problem in the first place. Because whenever our users open a mashup in the service, that seems to trigger the immediately launching of these "accidental" mashups. It is not clear how to avoid the behavior, and we are often forced to be reactive rather than proactive.