Open vfouqueron opened 1 month ago
Hi there! Can you try this one? https://github.com/rancher/rke2/issues/6307#issuecomment-2223536624
Thank you @dfaltum for your answer !
If I'm not mistaken port 4789 is for Windows nodes (as described there : https://docs.rke2.io/install/requirements#inbound-network-rules). There is currently no firewall on nodes, but there is also nothing listening on port 4789on the 192.168.9.30 at least. However, there is a service listening on port 8472 as expected and reachable from other nodes.
I runned /usr/sbin/ethtool -K flannel.1 tx-checksum-ip-generic off
but pods are still unreachable from other hosts
Environmental Info: RKE2 Version: rke2 version v1.28.12+rke2r1 (27989db0c83739c9f560e0567a3bf76177947122) go version go1.22.5 X:boringcrypto
Node(s) CPU architecture, OS, and Version: Linux VMALLKUB03 6.8.0-40-generic #40-Ubuntu SMP PREEMPT_DYNAMIC Fri Jul 5 10:34:03 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux
Cluster Configuration: I have only 3 servers, no agents
Describe the bug: The pods in node-1 and node-2 are reachable from all 3-nodes, but pods in node-3 are not reachable from node-1 and node-2.
I test this by doing pings to pods in the different nodes.
Steps To Reproduce: Here is my configurations :
Node 1
Node 2
Node 3
Expected behavior: All pods are reachable from all nodes
Actual behavior: Pods on node-3 are not reachable
Additional context / logs:
On node 1 :
On node 2 :
On node 3 :
rke-canal logs :