I have heard (mostly on reddit) that EBS seems to not fail but give very high IO delays upto 6 seconds. Is there a way in linux to quickly check the disk IO problems and get a figure?
I am thinking if I can script a test to check health and use IO problems as a way of terminating nodes automatically.
suggestions, help, idea, sysops jokes etc much appreciated
