This documentation is for the Latest version of Knative serving, which uses fleets and Anthos Service Mesh. Learn more.

The past version (Cloud Run for Anthos) has been archived but the documentation remains available for existing users.

Available versions

Latest
Archive

Migrate your workload traffic to Cloud Service Mesh

This step in the process allows you to choose how to rollout your changes. You can specify an initial amount of traffic that you want migrate to Cloud Service Mesh and then gradually increase the rollout to 100%.

Before you begin

You must first review and ensure that you meet the requirements for upgrading.
The commands on this page use the environment variables and migration script from the upgrade preparation step.
It is important that you perform each step of this process in the specified order.

Migrate all workload traffic to Cloud Service Mesh

In this section, the migration script is used to migrate your existing all traffic to Cloud Service Mesh.

Run the following command to set the percent of traffic you want migrated:
```
./migration-addon.sh --command migrate-traffic TRAFFIC_PERCENTAGE
```
Replace TRAFFIC_PERCENTAGE with the percent of traffic that you want to send to the to Cloud Service Mesh ingress. Valid values are: 0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100.

Note that if you have any issues with this step or want to perform a gradual rollout, you can re-run this migration script command multiple times until you reach 100%.

For example, if you choose to migrate 10% of traffic, the terminal output during the process looks similar to the following:
```
In this step, 10% of traffic from the previously installed
version of Istio is migrated to the Cloud Service Mesh ingress.

Setting the number of replicas of Istio ingress to 9,
and the number of replicas of Cloud Service Mesh ingress to 1
deployment.apps/istio-ingress patched

Waiting for all istio-ingress pods to be ready

All istio-ingress pods are ready.
deployment.apps/istio-ingressgateway patched

Waiting for all istio-ingressgateway pods to be ready

All istio-ingressgateway pods are ready.

10% traffic has been migrated to Cloud Service Mesh ingress
```
Note:
The total number of Cloud Service Mesh ingress pods and Istio ingress pods are scaled to 10 even if you currently have less ingress pods.
You can run the following command to view the status of the gke-system.
```
kubectl describe pod -n gke-system
```
If you receive the FailedScheduling warning, your cluster might not have sufficient resources to scale to the necessary number of ingress pods. To increase the number of nodes for your cluster and therefore, increase the amount of available resources, see Resizing a cluster in the Google Kubernetes Engine documentation.
You can monitor your traffic of your services, in the Google Cloud console:

Go to Knative serving

What's next

Finalize the upgrade and perform cleanup tasks

Rollback

If you need to rollback the changes that you made through the steps above, you can use the migration script to set the percentage of traffic to the Cloud Service Mesh ingress back to 0:

./migration-addon.sh --command migrate-traffic 0