msg="Could not retrieve CanonizedHost, rejecting "

time="2020-04-22T12:59:29-07:00" level=warning msg="Could not retrieve CanonizedHost, rejecting ".

What does this message mean, should I be concerned?

thanks
Brad

Hi Daxcor,

Are you using a LB with health checks by any chance? I've encountered this with F5 Health Monitor that sends HTTP/1.0 requests without a host header. The accompanied URL to that message should say if the request was done with HTTP/1.0 or not. To resolve this can you either add a host header to the HTTP/1.0 request or upgrade to HTTP/1.1. This is how I've solved this.

I have the same problem, the LoadBalancer I'm using is from Scaleway and they don't allow me to add headers. Any hint on a different LoadBalancer that can be used to grant ha to our kubernetes cluster on prem?

1 Like

Hello all, the message received is due to an incoming request not containing a Host header, undoubtedly due to being HTTP/1.0.

This is not a concern, and can be ignored safely. There is a PR to remove this warning for HTTP/1.9 connections:Ignore http 1.0 request host missing errors by dtomcej · Pull Request #8252 · traefik/traefik · GitHub

Hello @sandroden, if you are looking for on-prem LBs, you should look at MetalLB: https://metallb.universe.tf/

Hi @daniel.tomcej

this is reported by someone, my experience is different as reported here

If I'm correct that would not help in implementing High Availability. I think an external LaodBalancer that checks which nodes are reachable can give me more resilience. I have nodes in 2 different datacenters, on the contrary the entry point would not be under HA with metallb.