This message was deleted.
# rke2a
adamant-kite-43734
05/15/2025, 10:41 AMThis message was deleted.
c
creamy-pencil-82913
05/15/2025, 4:32 PMit doesn’t change the dns server address… it just runs a local cache pod and adds iptables rules that redirect dns traffic to that pod. this is covered in the docs.
https://docs.rke2.io/networking/networking_services#nodelocal-dnscache
c
calm-intern-77428
05/19/2025, 4:15 AMOoohhh okay, so it does not look the same as RKE1, got it. Thanks
m
miniature-rain-40358
08/04/2025, 10:49 AMThis thing fixed my dns issue. I was always getting timedout and this fixed it. Thanks. More like a mandatory stuff for me than performance one.
c
creamy-pencil-82913
08/04/2025, 3:58 PMIt shouldn't be mandatory. If DNS doesn't work when you don't have a replica on all nodes, then something is causing CNI overlay traffic between nodes to get dropped. You should find out why and fix that.
m
miniature-rain-40358
08/04/2025, 9:42 PMI think so, it works some times, but often fails. Now I am back to square one. It wont work.
I will study how to check the CNI overlay traffic. When I bash into coredns pods, the nslookup, curl works fine. But when I bash into other pods, like fleet related ones, nslookup works rarely or very slow.
Currently I thought I had this issue and I was working on that https://github.com/k3s-io/k3s/issues/5013
miniature-rain-40358
08/04/2025, 9:43 PMI am using RKE2, but saw someone using this to resolve dns issue.
My config is pretty basic.
# Listener# IPv4 address that apiserver uses to advertise to cluster membersadvertise-address: $(hostname -I | awk '{print $1}')# more hostnames/IPv4s as subject alternative names on the server TLS certtls-san:- $(hostname -I | awk '{print $1}')- $(wget -qO- ifconfig.me | grep "ip_addr:" | awk '{print $2}')tls-san-security: true# Networkingcni:- multus- canal# # cluster settingsservice-cidr: 172.16.0.0/16cluster-cidr: 172.170.0/16cluster-dns: 172.16.0.10cluster-domain: "cluster.local"# Kube clientwrite-kubeconfig-mode: "0644"write-kubeconfig: "/home/ubuntu/.kube/config"# clustertoken: SECRET_TOKEN# etcd settingsetcd-expose-metrics: trueetcd-snapshot-schedule-cron: "0 */23 * * *"etcd-snapshot-retention: 5etcd-snapshot-dir: /var/lib/rancher/rke2/db/snapshots# Node (details)node-name: master-$(hostname -I | awk '{print $1}')# Registering and starting kubelet with set of labelsnode-label:- "node-role=server"- "env=stag"# # Componentsdisable-cloud-controller: true19 Views