Stay organized with collections
Save and categorize content based on your preferences.
Inject sidecar proxies with Cloud Service Mesh
This document covers how to configure sidecar proxy injection with Cloud Service Mesh
to enhance network security, reliability, and observability. These functions are
abstracted away from the application's primary container and implemented in a
common out-of-process proxy (the sidecar), delivered as a separate container in
the same Pod. This provides the
Cloud Service Mesh's features without redesigning your
production applications to participate in a service mesh.
Automatic sidecar proxy injection (auto-injection) occurs when Cloud Service Mesh
detects a namespace label you configure for the workload's Pod. The proxy
intercepts all inbound and outbound traffic to the workloads and communicates
with Cloud Service Mesh.
Permissions required for these tasks
To perform the tasks on this page, you must have the
roles/container.clusterAdmin or a higher role. See
Google Kubernetes Engine roles for
details on the permissions included in this role.
Enabling automatic sidecar injection
The recommended way to inject sidecar proxies is to use the webhooks-based
automatic sidecar injector, although you can manually update your Pods'
Kubernetes configuration.
To enable auto-injection, you label your namespaces with the
default injection labels
if the default tag is set up, or with the
revision label to your namespace.
The label that you add also depends on whether you deployed
managed Cloud Service Mesh (with the
fleet API or with
asmcli), or
installed the in-cluster control plane. The label is used by the sidecar
injector webhook to associate injected sidecars with a particular control plane
revision.
To enable auto-injection:
In-cluster
Use the following command to locate the revision label on istiod:
kubectl -n istio-system get pods -l app=istiod --show-labels
In the output, under the LABELS column, note the value of the istiod
revision label, which follows the prefix istio.io/rev=. In this
example, the value is asm-11910-9.
Apply the revision label to namespaces and remove the istio-injection label
(if it exists). In the following command, NAMESPACE is
the name of the namespace where you want to enable auto-injection, and
REVISION is the revision label you noted in the
previous step.
You can ignore the message "istio-injection not found" in the
output. That means that the namespace didn't previously have the
istio-injection label, which you should expect in new
installations of Cloud Service Mesh or new deployments. Because auto-injection
behavior is undefined when a namespace has both the istio-injection
and the revision label, all kubectl label commands in the
Cloud Service Mesh documentation explicitly ensure that only one is set.
Restart the affected pods, using the steps in the next section.
Managed service mesh
Use the following command to locate the available release channels:
kubectl -n istio-system get controlplanerevision
The output is similar to the following:
NAME AGE
asm-managed 6d7h
In the output, select the value under the NAME column is the
REVISION label that corresponds to the available
release channel
for the Cloud Service Mesh version. Apply this label to your namespaces, and
remove the istio-injection label (if it exists).
In the following command, replace REVISION with the
revision label you noted above, and replace
NAMESPACE with the name of the namespace where you
want to enable auto-injection:
You can ignore the message "istio-injection not found" in the
output. That means that the namespace didn't previously have the
istio-injection label, which you should expect in new
installations of Cloud Service Mesh or new deployments. Because auto-injection
behavior is undefined when a namespace has both the istio-injection
and the revision label, all kubectl label commands in the
Cloud Service Mesh documentation explicitly ensure that only one is set.
Restart the affected pods, using the steps in the next section.
If you also deployed the optional
Google-managed data plane,
annotate the demo namespace as follows:
If you didn't use a Deployment, delete the Pods, and they are automatically
recreated with sidecars:
kubectl delete pod -n YOUR_NAMESPACE --all
Check that all the Pods in the namespace have sidecars injected:
kubectl get pod -n YOUR_NAMESPACE
In the following example output from the previous command, notice that the
READY column indicates there are two containers for each of your
workloads: the primary container and the container for the sidecar proxy.
NAME READY STATUS RESTARTS AGE
YOUR_WORKLOAD 2/2 Running 0 20s
...
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Hard to understand","hardToUnderstand","thumb-down"],["Incorrect information or sample code","incorrectInformationOrSampleCode","thumb-down"],["Missing the information/samples I need","missingTheInformationSamplesINeed","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2025-08-28 UTC."],[],[],null,["# Inject sidecar proxies with Cloud Service Mesh\n==============================================\n\nThis document covers how to configure sidecar proxy injection with Cloud Service Mesh\nto enhance network security, reliability, and observability. These functions are\nabstracted away from the application's primary container and implemented in a\ncommon out-of-process proxy (the sidecar), delivered as a separate container in\nthe same Pod. This provides the\n[Cloud Service Mesh's features](/service-mesh/v1.19/docs/overview) without redesigning your\nproduction applications to participate in a service mesh.\n\nAutomatic sidecar proxy injection (auto-injection) occurs when Cloud Service Mesh\ndetects a namespace label you configure for the workload's Pod. The proxy\nintercepts all inbound and outbound traffic to the workloads and communicates\nwith Cloud Service Mesh.\n\n#### Permissions required for these tasks\n\n\nTo perform the tasks on this page, you must have the\n`roles/container.clusterAdmin` or a higher role. See\n[Google Kubernetes Engine roles](/iam/docs/understanding-roles#container) for\ndetails on the permissions included in this role.\n\nEnabling automatic sidecar injection\n------------------------------------\n\nThe recommended way to inject sidecar proxies is to use the webhooks-based\nautomatic sidecar injector, although you can manually update your Pods'\nKubernetes configuration.\n\nTo enable auto-injection, you label your namespaces with the\n*[default injection labels](https://istio.io/latest/docs/setup/upgrade/canary/#default-tag)*\nif the default tag is set up, or with the\n[revision label](/service-mesh/v1.19/docs/revisions-overview) to your namespace.\nThe label that you add also depends on whether you deployed\nmanaged Cloud Service Mesh (with the\n[fleet API](/service-mesh/v1.19/docs/managed/provision-managed-anthos-service-mesh) or with\n[`asmcli`](/service-mesh/v1.19/docs/managed/provision-managed-anthos-service-mesh-asmcli)), or\ninstalled the in-cluster control plane. The label is used by the sidecar\ninjector webhook to associate injected sidecars with a particular control plane\nrevision.\n\nTo enable auto-injection: \n\n### In-cluster\n\n1. Use the following command to locate the revision label on `istiod`:\n\n kubectl -n istio-system get pods -l app=istiod --show-labels\n\n The output looks similar to the following: \n\n ```bash\n NAME READY STATUS RESTARTS AGE LABELS\n istiod-asm-11910-9-5788d57586-bljj4 1/1 Running 0 23h app=istiod,istio.io/rev=asm-11910-9,istio=istiod,pod-template-hash=5788d57586\n istiod-asm-11910-9-5788d57586-vsklm 1/1 Running 1 23h app=istiod,istio.io/rev=asm-11910-9,istio=istiod,pod-template-hash=5788d57586\n ```\n\n In the output, under the `LABELS` column, note the value of the `istiod`\n revision label, which follows the prefix `istio.io/rev=`. In this\n example, the value is `asm-11910-9`.\n\n\n | **Note:** You can substitute `istio.io/rev` with the `istio-injection=enabled` label if the [default tag](https://istio.io/latest/docs/setup/upgrade/canary/#default-tag) is configured. Verify the default tag exists by running ` istioctl tag list` with the `istioctl` from \u003cvar translate=\"no\"\u003eOUTPUT_DIR\u003c/var\u003e.\n\n \u003cbr /\u003e\n\n2. Apply the revision label to namespaces and remove the istio-injection label\n (if it exists). In the following command, \u003cvar translate=\"no\"\u003eNAMESPACE\u003c/var\u003e is\n the name of the namespace where you want to enable auto-injection, and\n \u003cvar translate=\"no\"\u003eREVISION\u003c/var\u003e is the revision label you noted in the\n previous step.\n\n kubectl label namespace \u003cvar translate=\"no\"\u003eNAMESPACE\u003c/var\u003e istio-injection- istio.io/rev=\u003cvar translate=\"no\"\u003eREVISION\u003c/var\u003e --overwrite\n\n\n You can ignore the message `\"istio-injection not found\"` in the\n output. That means that the namespace didn't previously have the\n `istio-injection` label, which you should expect in new\n installations of Cloud Service Mesh or new deployments. Because auto-injection\n behavior is undefined when a namespace has both the `istio-injection`\n and the revision label, all `kubectl label` commands in the\n Cloud Service Mesh documentation explicitly ensure that only one is set.\n3. Restart the affected pods, using the steps in the next section.\n\n### Managed service mesh\n\n1. Use the following command to locate the available release channels:\n\n kubectl -n istio-system get controlplanerevision\n\n The output is similar to the following: \n\n NAME AGE\n asm-managed 6d7h\n\n In the output, select the value under the `NAME` column is the\n \u003cvar translate=\"no\"\u003eREVISION\u003c/var\u003e label that corresponds to the available\n [release channel](/service-mesh/v1.19/docs/managed/select-a-release-channel#anthos_service_mesh_versions_per_channel)\n for the Cloud Service Mesh version. Apply this label to your namespaces, and\n remove the `istio-injection` label (if it exists).\n In the following command, replace \u003cvar translate=\"no\"\u003eREVISION\u003c/var\u003e with the\n revision label you noted above, and replace\n \u003cvar translate=\"no\"\u003eNAMESPACE\u003c/var\u003e` ` with the name of the namespace where you\n want to enable auto-injection: \n\n kubectl label namespace \u003cvar translate=\"no\"\u003eNAMESPACE\u003c/var\u003e istio-injection- istio.io/rev=\u003cvar translate=\"no\"\u003eREVISION\u003c/var\u003e --overwrite\n\n\n You can ignore the message `\"istio-injection not found\"` in the\n output. That means that the namespace didn't previously have the\n `istio-injection` label, which you should expect in new\n installations of Cloud Service Mesh or new deployments. Because auto-injection\n behavior is undefined when a namespace has both the `istio-injection`\n and the revision label, all `kubectl label` commands in the\n Cloud Service Mesh documentation explicitly ensure that only one is set.\n2. Restart the affected pods, using the steps in the next section.\n\n3. If you also deployed the optional\n [Google-managed data plane](/service-mesh/v1.19/docs/managed/provision-managed-anthos-service-mesh#managed-data-plane),\n annotate the `demo` namespace as follows:\n\n kubectl annotate --overwrite namespace \u003cvar translate=\"no\"\u003eYOUR_NAMESPACE\u003c/var\u003e \\\n mesh.cloud.google.com/proxy='{\"managed\":\"true\"}'\n\n### Restart Pods to update sidecar proxies\n\n| **Warning:** Unless you have a load balancer or router setup for [blue-green deployments](https://martinfowler.com/bliki/BlueGreenDeployment.html), make sure you test restarting Pods in a staging environment to verify that your services can handle any potential traffic interruption.\n\nWith automatic sidecar injection, you can update the sidecars for existing Pods\nwith a Pod restart:\n\nHow you restart Pods depends on if they were created as part of a\n[Deployment](https://kubernetes.io/docs/concepts/workloads/controllers/deployment/).\n\n1. If you used a Deployment, restart the Deployment, which restarts all Pods\n with sidecars:\n\n ```\n kubectl rollout restart deployment -n YOUR_NAMESPACE\n ```\n\n If you didn't use a Deployment, delete the Pods, and they are automatically\n recreated with sidecars: \n\n ```\n kubectl delete pod -n YOUR_NAMESPACE --all\n ```\n2. Check that all the Pods in the namespace have sidecars injected:\n\n ```\n kubectl get pod -n YOUR_NAMESPACE\n ```\n\n In the following example output from the previous command, notice that the\n `READY` column indicates there are two containers for each of your\n workloads: the primary container and the container for the sidecar proxy. \n\n ```\n NAME READY STATUS RESTARTS AGE\n YOUR_WORKLOAD 2/2 Running 0 20s\n ...\n ```\n\nWhat's next\n-----------\n\nLearn more about:\n\n- [Cloud Service Mesh control plane revisions](/service-mesh/v1.19/docs/revisions-overview)\n- [Deploying workloads](/kubernetes-engine/docs/how-to/deploying-workloads-overview)\n- [Customizing injection](https://istio.io/latest/docs/setup/additional-setup/sidecar-injection/#customizing-injection)"]]