Master Node Fails to Converge with Error "client: etcd cluster is unavailable or misconfigured

Problem

BareOS Platform9 Managed Kubernetes Cluster failed to converge with the following errors observed on one of the Master Node.

Error:  client: etcd cluster is unavailable or misconfigurederror #0: client: endpoint http://127.0.0.1:2379 exceeded header timeouterror #1: client: endpoint http://127.0.0.1:4001 exceeded header timeout

rafthttp: failed to dial fced1bcbbd9545a2 on stream Message (dial tcp 10.25.41.82:2380: getsockopt: no route to host)rafthttp: failed to dial fced1bcbbd9545a2 on stream MsgApp v2 (dial tcp 10.25.41.82:2380: getsockopt: no route to host)

Environment

• Platform9 Managed Kubernetes - All Versions

Cause

Cluster had its VIP address configured to be the same as that of one of the Master Nodes which resulted in conflict for the cluster to converge.

Resolution

1. Recreate the cluster with a VIP address separate from all the nodes within the cluster.
2. Refer to the following article which provides detailed steps on how to create a Multi Master BareOS Cluster on Platform9 Managed Kubernetes.

Additional Information