ENVIRONMENT:
I am running RKE2
Rancher's Kubernetes distribution. (docs)
I have a single node cluster.
CNI: Calico; Calico is using iptables
backend
iptables version: iptables v1.8.7 (legacy)
firewalld version: 0.9.3 (uses nftables
by default)
Issue:
I have a Hashicorp vault and Kubernetes cronjob running responsible for unsealing and initializing the vault on K8s cluster.
Because of some issue in network, the job is unable to communicate with the vault service running:
Error log from the job
OSError: [ErrNo 113] No route to host
Vault pod ip:
vault-0 10.42.108.2
ip route output
default via 10.225.247.129 dev eth0 proto dhcp
10.42.108.1 dev cali232527ff5fc scope link
10.42.108.2 dev calib56e549a1b4 scope link
10.42.108.3 dev cali7b4b3947ed0 scope link
10.42.108.4 dev cali480aa60fbbe scope link
10.42.108.6 dev cali46be8d885e0 scope link
10.42.108.7 dev cali0c60e5d7209 scope link
10.42.108.8 dev cali518c430520b scope link
10.42.108.9 dev cali505dfc32142 scope link
10.42.108.59 dev cali3ab04dcea6f scope link
10.42.108.63 dev cali5bcf89d2f73 scope link
10.225.247.128/25 dev eth0 proto kernel scope link src 10.225.247.171
So, veth calib56e549a1b4
entry corresponding to the pod IP is present in the table
I found few related docs on internet and followed the suggestions but nothing sees to be working.
ex: https://github.com/projectcalico/calico/issues/2322
I also tried changing firewalld
back end to iptable
, still no luck
Calico pods are all working fine.
I am unable to figure out why communication is failing. Any pointers or suggestions to debug this situation.
Feel free to ask if you need more info.
Thanks