Instructions on how to upgrade charts and images
Change-Id: I4006d620f9122081ea44ad8547b5ba615d337ac2
diff --git a/profiles/seba/operate/k8s-helm-ops.md b/profiles/seba/operate/k8s-helm-ops.md
index d122edd..ce35330 100644
--- a/profiles/seba/operate/k8s-helm-ops.md
+++ b/profiles/seba/operate/k8s-helm-ops.md
@@ -1 +1,76 @@
# Container Operations
+
+CORD uses Helm to manage the lifecycles of its component micro-services.
+This page describes how to perform some basic tasks using Helm commands.
+It assumes that charts are being installed from CORD's online Helm chart
+repository. To add this repository to your local Helm installation:
+
+```bash
+helm repo add cord https://charts.opencord.org
+helm repo update
+```
+
+If you have checked out the `helm-charts` repository locally and are installing
+charts from that, replace `cord/` in the examples below with the path to the
+chart.
+
+## Upgrade a Chart
+
+Upgrading a Helm chart will upgrade its services to the versions specified
+in the chart. For example, to upgrade the `seba-services` Helm chart to
+v1.0.3, assuming that has been published to the online chart repository:
+
+```bash
+helm upgrade --version=1.0.3 --reuse-values \
+ seba-services cord/seba-services
+```
+
+The `--reuse-values` flag instructs Helm to pass in the same values
+used when installing the original chart. If you want to specify a different
+set of values, you can omit this and instead specify new values using
+`-f myvalues.yaml`.
+
+## Upgrade a Single XOS Service's Image
+
+In order to upgrade a running micro-service to a new version, it's necessary to
+update the Helm chart used to launch the service to specify the new Docker
+image repository and/or tag. One way to do this is by upgrading the
+chart that was initially used to launch the service as mentioned above.
+However some charts (like `seba-services`) install multiple sub-charts,
+each of which installs its own XOS service. How can an operator upgrade a
+single service?
+
+An easy way to do this is with Helm's `--set` argument. To do this it's
+necessary to understand how the sub-chart specifies its image name and tag,
+as well as how the charts are nested. Using `seba-services` as an example,
+suppose that we have installed `seba-services` v1.0.2, and we want to upgrade
+the `fabric-crossconnect` service it installs from v1.1.4 to v1.1.5 without
+upgrading the `seba-services` chart to a new version. We can do so with this
+command:
+
+```bash
+helm upgrade --version=1.0.2 --reuse-values \
+ --set fabric-crossconnect.image.tag=1.1.5 \
+ seba-services cord/seba-services
+```
+
+The `seba-services` chart specifies the `fabric-crossconnect` chart
+as a requirement. The `fabric-crossconnect` chart's `values.yaml`
+file uses `image.tag` to specify the Docker image tag. So this command
+will pull down the new image and re-launch the `fabric-crossconnect`
+Kubernetes pod using this image.
+
+As another example, v1.0.0 of the `att-workflow` chart runs
+the `xosproject/att-workflow-driver:1.0.12` Docker image from Docker Hub. Suppose that you want to install image `myrepo/att-workflow-driver:test-image` in its place. To upgrade to this new image run:
+
+```bash
+helm upgrade --version=1.0.0 --reuse-values \
+ --set att-workflow-driver.image.repository=myrepo/att-workflow-driver
+ --set att-workflow-driver.image.tag=test-image \
+ att-workflow cord/att-workflow
+```
+
+To verify that you're specifying the `--set` arguments correctly, you can
+replace `helm upgrade` with `helm template` in the above commands. This will
+print out all the Kubernetes resources that Helm generates, and you can
+check that the image has actually been updated in the resources.