The "have you tried turning it off and on again" of DevOps. It makes a surprising amount of sense though, as long as your service is truly stateless, the restart can be easily orchestrated, and it results in no difference in operational costs.

If it's stateless, then why does rebuilding it change anything about the frequency of bugs popping up?

Ooh I can answer this one: because ask people if their container root is writeable, and get amused at the blank stares you get back.

I am currently fighting an ongoing battle at work to point out that the plans for our Mesos cluster have not factored in that the first outage we have will be when someone fills up the 100gb OS SSD because no one's given any thought to where the ephemeral container data goes.

I am a layman to devops. By "ephemeral container data" do you mean temporary files created by the service, temporary files created by the OS / other applications, or something else?

if both the code and the infra it's running on is stateless, then yeah.

So we're replacing somewhat not fully stable VMs running on somewhat not fully stable virtualization infrastructure with theoretically stable containers running on violently unstable container infrastructure?

Pretty much, yes.

Mutability is the root of all computing evils

And the source of all computing value.