Downgrade updates after testing

projectcalico · Dec 21, 2017 · 48d6343 · 48d6343
1 parent ac0e67d
commit 48d6343
Showing 1 changed file with 53 additions and 23 deletions.
diff --git a/master/getting-started/kubernetes/upgrade/downgrade.md b/master/getting-started/kubernetes/upgrade/downgrade.md
@@ -3,19 +3,40 @@ title: Downgrading Calico
 no_canonical: true
 ---
 
+## About downgrading {{site.prodname}}
+
 Under some circumstances, you may need to perform a downgrade and return your
-cluster to the previous version of {{site.prodname}}. You may need to do this
-before running `calico-upgrade complete` or afterwards. If you need to downgrade 
-your cluster after running `calico-upgrade complete`, you should do so as soon
-as possible to avoid an outage. Any pods created after `calico-upgrade complete`
-and before downgrading will lose networking.
+cluster to the previous version of {{site.prodname}}. If you need to downgrade
+you should do so as soon as possible to avoid an outage.
+
+The downgrade procedure varies according to how you originally installed
+{{site.prodname}} and your datastore type.
+
+- [Downgrading a self-hosted installation that uses the etcd
+  datastore](#downgrading-a-self-hosted-installation-that-uses-theetcd-datastore)
+  is necessary when migration has been done and the {{site.prodname}}
+  components have been upgraded.
+  > **Important**: Any pods created after `calico-upgrade complete` and
+  > before downgrading will lose networking.
+  {: .alert .alert-danger}
+
+- [Aborting the upgrade when using the etcd
+  datastore](#aborting-the-upgrade-when-using-the-etcd-datastore) can be done
+  when the migration has been started but the {{site.prodname}} components
+  have not been upgraded.
 
-## Downgrading a self-hosted installation
+- [Downgrading a self-hosted installation that uses the Kubernetes API
+  datastore](downgrading-a-self-hosted-installation-that-uses-the-kubernetes-api-datastore)
+
+- [Downgrading a custom installation](downgrading-a-custom-installation)
+
+## Downgrading a self-hosted installation that uses the etcd datastore
 
 If you have upgraded {{site.prodname}} by deploying the latest manifest,
 follow the steps here to downgrade.
 
-1. Follow the steps in [Aborting the upgrade](#aborting-the-upgrade)
+1. Follow the steps in
+   [Aborting the upgrade](#aborting-the-upgrade-that-uses-the-etcd-datastore)
    before downgrading the {{site.prodname}} components.
 
 1. Use the following commands to initiate a downgrade of the {{site.prodname}} components.
@@ -25,26 +46,15 @@ follow the steps here to downgrade.
    kubectl rollout undo deployment/calico-kube-controllers -n kube-system
    ```
 
-1. Watch the status of the downgrade as follows.
+1. Watch the status of the downgrade as follows. When it reports complete and
+   successful, {{site.prodname}} is downgraded to the previous version.
 
    ```
-   watch kubectl get pods -n kube-system
+   kubectl rollout status deployment/calico-kube-controllers -n kube-system
+   kubectl rollout status ds/calico-node -n kube-system
    ```
-
-   Verify that the status of all {{site.prodname}} pods indicate `Running`.
 
-   ```
-   calico-kube-controllers-6d4b9d6b5b-wlkfj   1/1       Running   0          3m
-   calico-node-hvvg8                          2/2       Running   0          3m
-   calico-node-vm8kh                          2/2       Running   0          3m
-   calico-node-w92wk                          2/2       Running   0          3m
-   ```
-
-## Downgrading a custom installation
-
-_Docs for this coming soon!_
-
-## Aborting the upgrade
+## Aborting the upgrade when using the etcd datastore
 
 If you have not upgraded the {{site.prodname}} components then follow
 the steps here to abort the upgrade and continue using your currently deployed
@@ -61,3 +71,23 @@ version of {{site.prodname}}.
 > **Important**: Do not use versions of `calicoctl` v3.0+ after aborting the upgrade.
 > Doing so may result in unexpected behavior and data.
 {: .alert .alert-danger}
+
+## Downgrading a self-hosted installation that uses the Kubernetes API datastore
+
+1. Use the following commands to initiate a downgrade of the {{site.prodname}} components.
+
+   ```
+   kubectl rollout undo ds/calico-node -n kube-system
+   ```
+
+1. Watch the status of the downgrade as follows. When it reports complete and
+   a successful, {{site.prodname}} is downgraded to the previous version.
+
+   ```
+   kubectl rollout status ds/calico-node -n kube-system
+   ```
+
+## Downgrading a custom installation
+
+_Docs for this coming soon!_
+