move node selection from --limit to --extra-vars=node<nodename>" #2948

qeqar · 2018-06-29T05:39:10Z

Hello,

while restricting remove-node.yml with --limit the pre and post task will not be executed!

  roles:
    - { role: remove-node/pre-remove, tags: pre-remove }

We don't have any valid host in group kube_master and --limit nodename
So on execution we get:

skipping: no hosts matched

So i propose we use --extra-vars for selection and the full inventory.

I changed the hosts: line in the node task and the selector in pre/post role task to "{{ node }}"

so we are able to call ansible-playbook -i <inventory> remove-node.yml --extra-vars "node=<nodename>"

Additionally it will not breack reset.yml. here we can use it with --limit there are no task which need diffrent group/host.

thanks in advanced,
mark

qeqar · 2018-06-29T05:52:48Z

Added some basic Doku

riverzhang · 2018-06-29T17:53:49Z

docs/getting-started.md

@@ -52,10 +52,12 @@ Remove nodes
 You may want to remove **worker** nodes to your existing cluster. This can be done by re-running the `remove-node.yml` playbook. First, all nodes will be drained, then stop some kubernetes services and delete some certificates, and finally execute the kubectl command to delete these nodes. This can be combined with the add node function, This is generally helpful when doing something like autoscaling your clusters. Of course if a node is not working, you can remove the node and install it again.

 - Add worker nodes to the list under kube-node if you want to delete them (or utilize a [dynamic inventory](https://docs.ansible.com/ansible/intro_dynamic_inventory.html)).


What the complete chapter?

Because we delete the change of node mode, we need to delete 54 lines.

riverzhang · 2018-06-29T17:56:34Z

docs/getting-started.md

@@ -52,10 +52,12 @@ Remove nodes
 You may want to remove **worker** nodes to your existing cluster. This can be done by re-running the `remove-node.yml` playbook. First, all nodes will be drained, then stop some kubernetes services and delete some certificates, and finally execute the kubectl command to delete these nodes. This can be combined with the add node function, This is generally helpful when doing something like autoscaling your clusters. Of course if a node is not working, you can remove the node and install it again.

 - Add worker nodes to the list under kube-node if you want to delete them (or utilize a [dynamic inventory](https://docs.ansible.com/ansible/intro_dynamic_inventory.html)).
+- Use `--extra-vars "node=<nodenname>"` to select the node you want to delete


node=node1,node2,node3,...... Is that like this？ when I delete nodes.

I'am not sure if this will work in the pre/post tasks.

For hosts: it is ok, but here

- - "{{ node }}"

i am not sure if ansible will make an array from a , seperated list.
But i will test it.

Ok added it

riverzhang · 2018-06-29T17:57:23Z

roles/remove-node/pre-remove/tasks/main.yml

@@ -9,7 +9,7 @@
      --timeout {{ drain_timeout }}
      --delete-local-data {{ item }}
  with_items:
-    - "{{ groups['kube-node'] }}"
+    - "{{ kube-node }}"


maybe {{ node }} ?

riverzhang · 2018-06-30T03:44:14Z

docs/getting-started.md

@@ -52,10 +52,12 @@ Remove nodes
 You may want to remove **worker** nodes to your existing cluster. This can be done by re-running the `remove-node.yml` playbook. First, all nodes will be drained, then stop some kubernetes services and delete some certificates, and finally execute the kubectl command to delete these nodes. This can be combined with the add node function, This is generally helpful when doing something like autoscaling your clusters. Of course if a node is not working, you can remove the node and install it again.

 - Add worker nodes to the list under kube-node if you want to delete them (or utilize a [dynamic inventory](https://docs.ansible.com/ansible/intro_dynamic_inventory.html)).
+- Use `--extra-vars "node=<nodename>,<nodename2>"` to select the node you want to delete


Add a "." at the end of the sentence.

riverzhang · 2018-07-02T09:40:02Z

cc @mattymo @ant31

mattymo · 2018-07-02T12:35:05Z

remove-node.yml

@@ -5,7 +5,7 @@
    ansible_ssh_pipelining: true
  gather_facts: true

- hosts: etcd:k8s-cluster:vault:calico-rr
+- hosts: "{{ node }}"


I am not opposed here. Let's default it though. Please change the line to:

hosts: "{{ node | default('etcd:k8s-cluster:vault:calico-rr') }}"

i can change this, but may i ask why? do you want to delete all nodes? or want to use --limit without per/post tasks?

because you should not break previous behavior, there're other users.

ok, change will come.

mattymo · 2018-07-02T12:35:12Z

remove-node.yml

@@ -22,7 +22,7 @@
  roles:
    - { role: remove-node/pre-remove, tags: pre-remove }

- hosts: kube-node
+- hosts: "{{ node }}"


same as above

mattymo · 2018-07-02T12:35:28Z

roles/remove-node/post-remove/tasks/main.yml

@@ -3,7 +3,7 @@
 - name: Delete node
  command: kubectl delete node {{ item }}
  with_items:
-    - "{{ groups['kube-node'] }}"
+    - "{{ node.split(',') }}"


this should default to members of kube-node group

mattymo · 2018-07-02T12:35:35Z

roles/remove-node/pre-remove/tasks/main.yml

@@ -9,7 +9,7 @@
      --timeout {{ drain_timeout }}
      --delete-local-data {{ item }}
  with_items:
-    - "{{ groups['kube-node'] }}"
+    - "{{ node.split(',') }}"


ant31 · 2018-07-02T14:57:17Z

We don't have any valid host in group kube_master

why ?

qeqar · 2018-07-02T15:20:26Z

@ant31 the --limit effects all hosts: definition. As we normally don't have any (or better the to be deleted) worker nodes in the kube_master group, the plays will be skipped.

So as long the node defined in --limit is not part of the kube_master group. it will never executed.

riverzhang · 2018-07-03T02:22:26Z

docs/getting-started.md

@@ -51,11 +51,12 @@ Remove nodes

 You may want to remove **worker** nodes to your existing cluster. This can be done by re-running the `remove-node.yml` playbook. First, all nodes will be drained, then stop some kubernetes services and delete some certificates, and finally execute the kubectl command to delete these nodes. This can be combined with the add node function, This is generally helpful when doing something like autoscaling your clusters. Of course if a node is not working, you can remove the node and install it again.

- Add worker nodes to the list under kube-node if you want to delete them (or utilize a [dynamic inventory](https://docs.ansible.com/ansible/intro_dynamic_inventory.html)).


The description of these two kinds of deletions should be added. There is only one description at the moment.

Add worker nodes to the list under kube-node if you want to delete them (or utilize a dynamic inventory).
ansible-playbook -i inventory/mycluster/hosts.ini remove-node.yml -b -v
--private-key=~/.ssh/private_key
We still need this description. Please add. Thanks.

riverzhang

LGTM

riverzhang reviewed Jun 29, 2018

View reviewed changes

qeqar force-pushed the remove-node-limit branch 5 times, most recently from ea80e02 to 16a46b3 Compare June 29, 2018 19:46

riverzhang reviewed Jun 30, 2018

View reviewed changes

qeqar force-pushed the remove-node-limit branch 2 times, most recently from 1163c0e to 18e3245 Compare June 30, 2018 11:50

mattymo reviewed Jul 2, 2018

View reviewed changes

move node selection from --limit to --extra-vars=node<nodename>"

b548f6f

qeqar force-pushed the remove-node-limit branch from 18e3245 to 8242216 Compare July 2, 2018 18:05

riverzhang reviewed Jul 3, 2018

View reviewed changes

qeqar force-pushed the remove-node-limit branch 2 times, most recently from d0f2ec4 to a1c19fd Compare July 3, 2018 08:37

[remove-node] add doku for nodeselector

af635ff

qeqar force-pushed the remove-node-limit branch from a1c19fd to af635ff Compare July 3, 2018 08:38

riverzhang approved these changes Jul 3, 2018

View reviewed changes

riverzhang merged commit 32a6ca4 into kubernetes-sigs:master Jul 3, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

move node selection from --limit to --extra-vars=node<nodename>" #2948

move node selection from --limit to --extra-vars=node<nodename>" #2948

qeqar commented Jun 29, 2018 •

edited

Loading

qeqar commented Jun 29, 2018

riverzhang Jun 29, 2018

qeqar Jun 29, 2018

riverzhang Jun 30, 2018

riverzhang Jun 29, 2018

qeqar Jun 29, 2018

qeqar Jun 29, 2018

riverzhang Jun 29, 2018

qeqar Jun 29, 2018

riverzhang Jun 30, 2018

riverzhang commented Jul 2, 2018

mattymo Jul 2, 2018

qeqar Jul 2, 2018

ant31 Jul 2, 2018

qeqar Jul 2, 2018

mattymo Jul 2, 2018

mattymo Jul 2, 2018

mattymo Jul 2, 2018

ant31 commented Jul 2, 2018

qeqar commented Jul 2, 2018

riverzhang Jul 3, 2018

riverzhang Jul 3, 2018

riverzhang left a comment

		@@ -52,10 +52,12 @@ Remove nodes
		You may want to remove worker nodes to your existing cluster. This can be done by re-running the `remove-node.yml` playbook. First, all nodes will be drained, then stop some kubernetes services and delete some certificates, and finally execute the kubectl command to delete these nodes. This can be combined with the add node function, This is generally helpful when doing something like autoscaling your clusters. Of course if a node is not working, you can remove the node and install it again.

		- Add worker nodes to the list under kube-node if you want to delete them (or utilize a [dynamic inventory](https://docs.ansible.com/ansible/intro_dynamic_inventory.html)).

		@@ -51,11 +51,12 @@ Remove nodes

		You may want to remove worker nodes to your existing cluster. This can be done by re-running the `remove-node.yml` playbook. First, all nodes will be drained, then stop some kubernetes services and delete some certificates, and finally execute the kubectl command to delete these nodes. This can be combined with the add node function, This is generally helpful when doing something like autoscaling your clusters. Of course if a node is not working, you can remove the node and install it again.

		- Add worker nodes to the list under kube-node if you want to delete them (or utilize a [dynamic inventory](https://docs.ansible.com/ansible/intro_dynamic_inventory.html)).

move node selection from --limit to --extra-vars=node<nodename>" #2948

move node selection from --limit to --extra-vars=node<nodename>" #2948

Conversation

qeqar commented Jun 29, 2018 • edited Loading

qeqar commented Jun 29, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

riverzhang commented Jul 2, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ant31 commented Jul 2, 2018

qeqar commented Jul 2, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

riverzhang left a comment

Choose a reason for hiding this comment

qeqar commented Jun 29, 2018 •

edited

Loading