I have an install of discourse based on docker that may drive me crazy.
As soon as I put some average load on the server (like doing a compilation), the run queue (observed by top) will rise to values of 10 and much more (even 100), and will render the VM (ESXi based) virtually inoperable.
ESXi Performance counters show that there is nearly no I/O, no memory shortage, but the CPU is constantly over-loaded.
Who is likely to be blamed: might it be docker integration on SLES?
Thanks for sharing your opinion or tips on how to track down the root cause of the issue.