Problem during genesis with calico-etcd-anchor and calico-etcd
Hello,
I've managed to install many kubernetes clusters using promenade, armada. But for some strange reasons the genesis process doesn't work in my current environment.
The only difference with my previous environment is that the interface I'm using is a bonded interface.
The genesis process get stucks with :
Error: unhealthy cluster
+ date +%s
+ now=1532799485
+ [ 1532799485 -gt 1532799726 ]
+ sleep 15
+ etcdctl member list
+ grep 192.168.41.244
56c1d8d3e228b500, started, signals01, https://192.168.41.244:6667, https://192.168.41.244:6666
+ ETCDCTL_ENDPOINTS=https://192.168.41.244:6666 etcdctl endpoint health
https://192.168.41.244:6666 is unhealthy: failed to connect: context deadline exceeded
Error: unhealthy cluster
+ date +%s
+ now=1532799503
+ [ 1532799503 -gt 1532799726 ]
+ sleep 15
Otherwise the other pods look fine:
NAMESPACE NAME READY STATUS RESTARTS AGE
kube-system auxiliary-etcd-signals01 3/3 Running 0 37m
kube-system bootstrap-armada-signals01 4/4 Running 0 38m
kube-system calico-etcd-anchor-5jqdk 0/1 Running 0 30m
kube-system calico-etcd-signals01 1/1 Running 0 2m
kube-system calico-kube-policy-controllers-54cd86cf9b-5hphl 1/1 Running 4 18m
kube-system calico-node-82xqk 2/2 Running 1 26m
kube-system calico-settings-v2r2w 0/1 Completed 0 26m
kube-system coredns-86d5687d78-gslbp 1/1 Running 0 15m
kube-system coredns-86d5687d78-jq2pc 1/1 Running 0 15m
kube-system haproxy-anchor-nlpc6 1/1 Running 0 5m
kube-system haproxy-signals01 1/1 Running 0 5m
kube-system kubernetes-apiserver-signals01 1/1 Running 0 37m
kube-system kubernetes-controller-manager-signals01 1/1 Running 0 38m
kube-system kubernetes-etcd-signals01 1/1 Running 0 37m
kube-system kubernetes-proxy-b9tlx 1/1 Running 0 37m
kube-system kubernetes-scheduler-signals01 1/1 Running 0 38m
Do you have an idea what could be wrong? Is it possible that bonding might have an impact? I've spend 4 days trying to debug it but no results...
Best, James
Hi,
I've managed to fix it, I think it was due to wrong default routes. So I changed the routes, disabled apparmor and took the latest kernel for Ubuntu Xenial and it works.
Best, James
On Sat, Jul 28, 2018 at 7:42 PM, James Devon jr8586335@gmail.com wrote:
Hello,
I've managed to install many kubernetes clusters using promenade, armada. But for some strange reasons the genesis process doesn't work in my current environment.
The only difference with my previous environment is that the interface I'm using is a bonded interface.
The genesis process get stucks with :
Error: unhealthy cluster
date +%s
now=1532799485
[ 1532799485 -gt 1532799726 ]
sleep 15
etcdctl member list
grep 192.168.41.244
56c1d8d3e228b500, started, signals01, https://192.168.41.244:6667, https://192.168.41.244:6666
- ETCDCTL_ENDPOINTS=https://192.168.41.244:6666 etcdctl endpoint health
https://192.168.41.244:6666 is unhealthy: failed to connect: context deadline exceeded
Error: unhealthy cluster
date +%s
now=1532799503
[ 1532799503 -gt 1532799726 ]
sleep 15
Otherwise the other pods look fine:
NAMESPACE NAME READY STATUS RESTARTS AGE
kube-system auxiliary-etcd-signals01 3/3 Running 0 37m
kube-system bootstrap-armada-signals01 4/4 Running 0 38m
kube-system calico-etcd-anchor-5jqdk 0/1 Running 0 30m
kube-system calico-etcd-signals01 1/1 Running 0 2m
kube-system calico-kube-policy-controllers-54cd86cf9b-5hphl 1/1 Running 4 18m
kube-system calico-node-82xqk 2/2 Running 1 26m
kube-system calico-settings-v2r2w 0/1 Completed 0 26m
kube-system coredns-86d5687d78-gslbp 1/1 Running 0 15m
kube-system coredns-86d5687d78-jq2pc 1/1 Running 0 15m
kube-system haproxy-anchor-nlpc6 1/1 Running 0 5m
kube-system haproxy-signals01 1/1 Running 0 5m
kube-system kubernetes-apiserver-signals01 1/1 Running 0 37m
kube-system kubernetes-controller-manager-signals01 1/1 Running 0 38m
kube-system kubernetes-etcd-signals01 1/1 Running 0 37m
kube-system kubernetes-proxy-b9tlx 1/1 Running 0 37m
kube-system kubernetes-scheduler-signals01 1/1 Running 0 38m
Do you have an idea what could be wrong? Is it possible that bonding might have an impact? I've spend 4 days trying to debug it but no results...
Best, James
participants (1)
-
James Devon