Node-based upgrade fails with: "Failed to template loop_control.label: 'None' has no attribute 'item'"

[renesepp]

Environment:

• Cloud provider or hardware configuration:
  Onprem VMware ESX
• OS (printf "$(uname -srm)\n$(cat /etc/os-release)\n"):

Linux 4.18.0-372.19.1.el8_6.x86_64 x86_64
NAME="Rocky Linux"
VERSION="8.6 (Green Obsidian)"

• Version of Ansible (ansible --version):
  ansible [core 2.11.11]
• Version of Python (python --version):
  Python 3.6.8

Kubespray version (commit) (git rev-parse --short HEAD):
c4346e590
Network plugin used:
Calico

Command used to invoke ansible:
ansible-playbook -i inventory/mycluster/hosts.yaml -b -e kube_version=v1.25.6 --become --become-user=root --ask-pass --ask become-pass upgrade-cluster.yml --limit "testk8-n3"
Output of ansible run:

TASK [kubernetes-apps/network_plugin/multus : Multus | Start resources] ************************************************************************************************************************************** failed: [testk8-n3 -> {{ groups['kube_control_plane'][0] }}] (item=None) => {"ansible_loop_var": "item", "changed": false, "item": null, "msg": "Failed to template loop_control.label: 'None' has no attribute 'item'", "skip_reason": "Conditional result was False"}
Anything else do we need to know:

Only happens when issuing --limit pattern(fine without --limit)
https://github.com/kubernetes-sigs/kubespray/blob/master/docs/upgrades.md#node-based-upgrade

[mvilche]

I have the same problem when updating a single node of the cluster.

TASK [kubernetes-apps/network_plugin/multus : Multus | Start resources] ********
failed: [kube-noprod-worker05 -> {{ groups['kube_control_plane'][0] }}] (item=None) => {"ansible_loop_var": "item", "changed": false, "item": null, "msg": "Failed to template loop_control.label: 'None' has no attribute 'item'", "skip_reason": "Conditional result was False"}

[splix]

I have the same issue when running with --limit. But seems to be working when I manually disable the plugin by running ansible with --skip-tags=multus

[splix]

It seems that with --limit it doesn't use the correct networking model. In my case it also had added the kube-proxy pods, and other things for a default model though I tried to setup with Cilium. When I run without a --limit it doesn't create such pods.

[ksyblast]

I have the same issue when trying to setup a multi-node cluster with calico, kube_network_plugin_multus: false , I do not use --limit.

ansible-playbook  -i inventory/myinv/hosts.yml   cluster.yml    -D  --become --extra-vars'=ansible_ssh_timeout=50'

Installed from kubespray tag v2.21.0

[xinnj]

Met the same problem. Have to use '--skip-tags=multus' to skip it for now.

[vyom-soft]

I have the same issue when trying to setup a multi-node cluster with calico, kube_network_plugin_multus: false , I do not use --limit.

ansible-playbook  -i inventory/myinv/hosts.yml   cluster.yml    -D  --become --extra-vars'=ansible_ssh_timeout=50'

Installed from kubespray tag v2.21.0

Were you able to execute the ansible-playbook successfully for multi-node cluster. Or you also used that flag --skip-tags=multus

[ksyblast]

I have the same issue when trying to setup a multi-node cluster with calico, kube_network_plugin_multus: false , I do not use --limit.

ansible-playbook  -i inventory/myinv/hosts.yml   cluster.yml    -D  --become --extra-vars'=ansible_ssh_timeout=50'

Installed from kubespray tag v2.21.0

Were you able to execute the ansible-playbook successfully for multi-node cluster. Or you also used that flag --skip-tags=multus

I also used --skip-tags=multus

[yankay]

HI @maxime1907

Would you please provide a PR to resolve it :-)

[behdadkh]

I am on branch 2.22 and still do get this issue!

TASK [kubernetes-apps/network_plugin/multus : Multus | Start resources] ************************************************************************************************************************************** failed: [de-small-2 -> {{ groups['kube_control_plane'][0] }}] (item=None) => {"ansible_loop_var": "item", "changed": false, "item": null, "msg": "Failed to template loop_control.label: 'None' has no attribute 'item'", "skip_reason": "Conditional result was False"}

[ayoubeddafali]

I can confirm that it still happens when upgrading only one node ( for some reason )

[MatthiasLohr]

Also still affected. No solution in sight?

[nicolas-goudry]

/assign