Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

    I felt like if anything went wrong I'd be in a world of hurt trying to debug it.
This is the part very few k8s aficionados mention: how complicated is it to troubleshoot when something goes wrong?


The core k8s, not that bad, once you've learned where to look - logs on pods, events, and kubelet logs from journalctl. That last one is important not to forget when debugging plugin controllers.


Not to mention when folks talk HA and you ask about federation they just give you a blank stare.

k8s is HA in the same region if you are using GKE. Not really what you can call HA


Google just posted a blog[0] about multi-region clusters and ingress.

It's a little hacky as you need to create multiple clusters in each zone/region and then with an anycast IP attached to a Google Cloud Load Balancer.

[0] https://cloudplatform.googleblog.com/2018/06/How-to-deploy-g...


Hacky is not production ready.


Its not hacky in that sense. They provide a new kubemci tool and I'm sure it will be given the k8 release treatment.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: