Stephan

@knicknic That is already doable if I ubderstood6you correctly. You can already prevent metallb from allocating (still validating) and can set to svc.status.loadBalancerIP yourself and have that picked up by...

Just out of curiousity: Why do you need node-specific peer-ips on your ToR switches? Is this a precondition in packet? In out on-premise setup, we peer with ToR switches as...

You can of course add a second service with a type of your choice, exposing the apiserver a second time. If your apiserver runs as a pod, you have a...

Jumping in here: We observe similar using FlatCar underneith. When typha fails, calico-node goes into a crash loop, not being able to update it's bird.conf. It would be great to...

> when typha fails This was meant as "all typha pods fail", e.g. because the k8s apiserver fails. This was more likely in some older version (last year) when a...

> I believe this is still possible with VXLAN - you would use VXLAN for pod-to-pod routing, and can still enable BGP for advertisement of service IPs. It doesn't fully...

I could reproduce hung task errors on bare metal by running `stress -i 10 -d 5` within a pod's emptyDir, backed by the host's SSD Raid1 after a while. I...

I have a similar issue: When I add data for the top right or bottom left quadrant with a radius of >=100, the JS fails. Maybe this is identical, I...

Current prerequisites of course are documented as just k8s 1.8+ - my proposal would only work for k8s 1.18+

IIRC and from @juliantaylor 's description it was advertised via BGP (see `100.70.221.192/26 via xx.xx.199.27 dev tunl0 proto bird onlink`) in the routes of other nodes, and it was a...