You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
What happened:
CNI plugin is failing to launch on our hosts with two ENI already. This is part of how we have our ETCD hosts setup, they boot up with a random eni, and then allocation a known ENI for consistent IP allocation.
Attach logs
From the logs of the init container:
~ ❯ ~/bin/k -n kube-system logs aws-node-kwpdh -c aws-vpc-cni-init
+ PLUGIN_BINS='loopback portmap bandwidth aws-cni-support.sh'
+ for b in '$PLUGIN_BINS'
+ '[' '!' -f loopback ']'
Copying CNI plugin binaries ...
+ for b in '$PLUGIN_BINS'
+ '[' '!' -f portmap ']'
+ for b in '$PLUGIN_BINS'
+ '[' '!' -f bandwidth ']'
+ for b in '$PLUGIN_BINS'
+ '[' '!' -f aws-cni-support.sh ']'
+ HOST_CNI_BIN_PATH=/host/opt/cni/bin
+ echo 'Copying CNI plugin binaries ... '
+ for b in '$PLUGIN_BINS'
+ install loopback /host/opt/cni/bin
+ for b in '$PLUGIN_BINS'
+ install portmap /host/opt/cni/bin
+ for b in '$PLUGIN_BINS'
+ install bandwidth /host/opt/cni/bin
+ for b in '$PLUGIN_BINS'
+ install aws-cni-support.sh /host/opt/cni/bin
+ echo 'Configure rp_filter loose... '
++ curl -X PUT http://169.254.169.254/latest/api/token -H 'X-aws-ec2-metadata-token-ttl-seconds: 60'
Configure rp_filter loose...
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 56 100 56 0 0 56000 0 --:--:-- --:--:-- --:--:-- 56000
+ TOKEN=AQAEAIRBUrXmr38hU9BstXIoL9aDWAWfZGe_7D9WB7wN-IaHU4i9Dw==
++ curl -H 'X-aws-ec2-metadata-token: AQAEAIRBUrXmr38hU9BstXIoL9aDWAWfZGe_7D9WB7wN-IaHU4i9Dw==' http://169.254.169.254/latest/meta-data/local-ipv4
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 11 100 11 0 0 11000 0 --:--:-- --:--:-- --:--:-- 11000
+ HOST_IP=10.21.4.253
++ ip -4 -o a
++ awk '{print $2}'
++ grep 10.21.4.253
+ PRIMARY_IF=
What you expected to happen:
CNI plugin runs and doesn't error out at start up
How to reproduce it (as minimally and precisely as possible):
Bring up a system with one ENI, then attach a new/replacement one and detach the original, then add the vpc-cni.
Anything else we need to know?:
Based on my manual testing, this appears to work with the latest regex committed to master in 5d05d33
I think we just need a new bug release cut.
What happened:
CNI plugin is failing to launch on our hosts with two ENI already. This is part of how we have our ETCD hosts setup, they boot up with a random eni, and then allocation a known ENI for consistent IP allocation.
Attach logs
From the logs of the init container:
From the pod describe:
What you expected to happen:
CNI plugin runs and doesn't error out at start up
How to reproduce it (as minimally and precisely as possible):
Bring up a system with one ENI, then attach a new/replacement one and detach the original, then add the vpc-cni.
Anything else we need to know?:
Based on my manual testing, this appears to work with the latest regex committed to master in 5d05d33
I think we just need a new bug release cut.
Environment:
Kubernetes: v1.17.5
CNI: 1.7.5
OS: Flatcar 2492.0.0
Kernel: 5.4.35-flatcar
kubectl version
):cat /etc/os-release
):uname -a
):The text was updated successfully, but these errors were encountered: