Merge pull request #6335 from kwoodson/node_groups_refactor

Node group management update.

2 files changed, 23 insertions, 16 deletions

diff --git a/playbooks/aws/openshift-cluster/accept.yml b/playbooks/aws/openshift-cluster/accept.yml
index cab2f1e40..e7bed4f6e 100755
--- a/playbooks/aws/openshift-cluster/accept.yml
+++ b/playbooks/aws/openshift-cluster/accept.yml
@@ -18,7 +18,7 @@
       name: lib_openshift
 
   - name: fetch masters
-    ec2_remote_facts:
+    ec2_instance_facts:
       region: "{{ openshift_aws_region | default('us-east-1') }}"
       filters:
         "tag:clusterid": "{{ openshift_aws_clusterid | default('default') }}"
@@ -30,7 +30,7 @@
     until: "'instances' in mastersout and mastersout.instances|length > 0"
 
   - name: fetch new node instances
-    ec2_remote_facts:
+    ec2_instance_facts:
       region: "{{ openshift_aws_region | default('us-east-1') }}"
       filters:
         "tag:clusterid": "{{ openshift_aws_clusterid | default('default') }}"
diff --git a/playbooks/common/openshift-cluster/upgrades/upgrade_scale_group.yml b/playbooks/common/openshift-cluster/upgrades/upgrade_scale_group.yml
index 47410dff3..4fc897a57 100644
--- a/playbooks/common/openshift-cluster/upgrades/upgrade_scale_group.yml
+++ b/playbooks/common/openshift-cluster/upgrades/upgrade_scale_group.yml
@@ -11,25 +11,19 @@
       msg: "Ensure that new scale groups were provisioned before proceeding to update."
     when:
     - "'oo_sg_new_nodes' not in groups or groups.oo_sg_new_nodes|length == 0"
+    - "'oo_sg_current_nodes' not in groups or groups.oo_sg_current_nodes|length == 0"
+    - groups.oo_sg_current_nodes == groups.oo_sg_new_nodes
 
 - name: initialize upgrade bits
   import_playbook: init.yml
 
-- name: Drain and upgrade nodes
+- name: unschedule nodes
   hosts: oo_sg_current_nodes
-  # This var must be set with -e on invocation, as it is not a per-host inventory var
-  # and is evaluated early. Values such as "20%" can also be used.
-  serial: "{{ openshift_upgrade_nodes_serial | default(1) }}"
-  max_fail_percentage: "{{ openshift_upgrade_nodes_max_fail_percentage | default(0) }}"
-
-  pre_tasks:
+  tasks:
   - name: Load lib_openshift modules
-    include_role:
+    import_role:
       name: ../roles/lib_openshift
 
-  # TODO: To better handle re-trying failed upgrades, it would be nice to check if the node
-  # or docker actually needs an upgrade before proceeding. Perhaps best to save this until
-  # we merge upgrade functionality into the base roles and a normal config.yml playbook run.
   - name: Mark node unschedulable
     oc_adm_manage_node:
       node: "{{ openshift.node.nodename | lower }}"
@@ -40,14 +34,27 @@
     register: node_unschedulable
     until: node_unschedulable|succeeded
 
+- name: Drain nodes
+  hosts: oo_sg_current_nodes
+  # This var must be set with -e on invocation, as it is not a per-host inventory var
+  # and is evaluated early. Values such as "20%" can also be used.
+  serial: "{{ openshift_upgrade_nodes_serial | default(1) }}"
+  max_fail_percentage: "{{ openshift_upgrade_nodes_max_fail_percentage | default(0) }}"
+  tasks:
   - name: Drain Node for Kubelet upgrade
     command: >
-      {{ hostvars[groups.oo_first_master.0].openshift.common.client_binary }} adm drain {{ openshift.node.nodename | lower }} --config={{ openshift.common.config_base }}/master/admin.kubeconfig --force --delete-local-data --ignore-daemonsets
+      {{ hostvars[groups.oo_first_master.0].openshift.common.client_binary }} adm drain {{ openshift.node.nodename | lower }}
+      --config={{ openshift.common.config_base }}/master/admin.kubeconfig
+      --force --delete-local-data --ignore-daemonsets
+      --timeout={{ openshift_upgrade_nodes_drain_timeout | default(0) }}s
     delegate_to: "{{ groups.oo_first_master.0 }}"
     register: l_upgrade_nodes_drain_result
     until: not l_upgrade_nodes_drain_result | failed
-    retries: 60
-    delay: 60
+    retries: "{{ 1 if openshift_upgrade_nodes_drain_timeout | default(0) == '0' else 0  | int }}"
+    delay: 5
+    failed_when:
+    - l_upgrade_nodes_drain_result | failed
+    - openshift_upgrade_nodes_drain_timeout | default(0) == '0'
 
 # Alright, let's clean up!
 - name: clean up the old scale group