Cephadm Revisited: Unveiling New Improvements and Enhanced Production Readiness

In this video, I revisit the topic of cephadm. Last time, I thought it was not ready for production, but I showed off some of the improvements made in the process. We talk about the different steps required in the console to get the cluster started, and then I look at the GUI for the setup process of setting up services, hosts, and OSDs. There are still a couple of rough edges, but it's much improved since the last time we looked into the topic.

Support the channel and donate a coffee:

Or visit my blog at:

Outro music: Sanaas Scylla

#cephadm #osd #services

[MrWadezz]

Damn just when I'm finally getting around to use cephadm you release an updated video
Thanks for that, always really interesting stuff !

[tqnpersonal]

Great video.

Is it possible to rerun the bootstrap command on the already bootstrapped node, if I want to use additional flags?

[claudiolengt]

You really make it look easy. Thanks for it.

Would you use containers in a cluster with 10 OSD servers with 36 disks each and 3 additional servers for Mon, Mngr, Rados, etc. services?
What type of installation would you do?

[christianpeters]

Great overview! 🙂Is it still the case that cephadmin installs all services in docker containers and doing a CLI will setup the services with systemd!? Is there any chance to debug the services installed by cephadmin in containers or would you still recommend the CLI way of deployment?

[Darkk]

I agree that cephadm needs some more polishing to get full use of CEPH. CLI will always be available for those who want to get under the hood.