4 ms·
>Something along the lines of what Go brought to the game. Where the biggest adopter, Kubernetes, is ironically a way too complex system of automation routines
by sascha_sl 8y ago
>Something along the lines of what Go brought to the game.
Where the biggest adopter, Kubernetes, is ironically a way too complex system of automation routines that still regularly fail in production environments. And oh dear, if they fail, it's gonna be hard to figure out what exactly went wrong. We uncovered a bug in the job controller just yesterday that was probably not biting anyone else due to how a race condition usually plays out, but this kind of bug is way too common.
- theredbox 8y agoYeah that's the irony of it. But Go is just a tool. People were building monstrosities with bash before and will be long after we build the mythical AI Cloud.
- apta 8y ago> But Go is just a tool. Yes, but there are better tools suited for the job. Ironically, or perhaps not, Java or .NET would have faired better due to maturity of debugging and introspection tools.
- nova22033 8y ago>Where the biggest adopter, Kubernetes, is ironically a way too complex system of automation routines that still regularly fail in production environments. regularly fail? Do you have any data to back that up? No, a google search for one kubernetes failure doesn't count as proof of "regular failure"
- sascha_sl 8y agoI'm in an SRE-like position for an otherwise unmanaged Kubernetes cluster. We have code merged in the repo as well as a few PRs still open. I should note that by regularly I mean edge cases and quirkiness, not that your run of the mill cluster mostly using deployments is going to randomly explode. But our teams still find ways to regularly break it in unexpected ways. So I guess you could call that anecdotal if you want.