node-feature-discovery VS Duplicacy

Oh, sorry about that. Please use this link instead: https://github.com/kubernetes-sigs/node-feature-discovery

Look into NFD as it will automatically maintain labels based on hardware attributes including accelerators, CPU models, CPU instruction sets and other hardware detail.

I installed this plugin in combination with Node Feature Discovery, so that the plugin only gets deployed on nodes with an iGPU.

node-feature-discovery and descheduler

.1. https://github.com/kubernetes-sigs/node-feature-discovery

When deploying without the helm chart, you need to define that in the ConfigMap. Looks like it’s here: https://github.com/kubernetes-sigs/node-feature-discovery/blob/master/nfd-daemonset-combined.yaml.template#L137-L229

I have been having great luck with incremental backups with the very similar named Duplicacy https://duplicacy.com/

Pro: No need for a special index database. The chunks are placed in the file system. This explains it in greater detail. Seems to place great emphasis on reliability, which is important for me. Versioning is also supported.

Having all your data in one place isn't wise though, so I am planning on storing encrypted backups on Dropbox and Backblaze B2 using Duplicity so that I am following the 3-2-1 backup rule.

I tried a bunch of different ways but ultimately settled on Duplicacy [0].

It runs inside a Docker container and backs up both my data as well as configurations like my docker compose file and smb.conf.

Off site storage was Backblaze B2, but I moved to Hetzner. Likely will move back just because B2 is cheaper and a bit faster for my region.

Another layer of backup I do is use Duplicacy to backup to a portable hard drive occasionally that I keep off site.

[0] https://duplicacy.com/

Duplicacy

I use Kopia and works well. Have also used this https://duplicacy.com