One of the main advantages of embracing containers is “lightweight virtualization.” Since each container is just a thin layer around the containerized processes, the user gains enormous efficiencies, for example by increasing the container density per host, or by spinning containers up and down at a very fast pace.
However, as the troubleshooting story in the article will show, this lightweight virtualization comes at the cost of sharing the underlying kernel among all containers, and in some circumstances, this can lead to surprising and undesirable effects that container users typically don’t think about.
This troubleshooting tale is rather involved. I’ve started from the basics and worked up to the more complex material in the hope that readers at all levels can get value out of it.
Read more at DZone