Google's primary support objective is to resolve production incidents as quickly as possible. We do this by understanding your configuration, analyzing logs and metrics, and collaborating with partners to solve incidents quickly.
Google Cloud offers a variety of support packages to accommodate your support needs. All Google Cloud Support packages include support for Anthos and Anthos on bare metal. If you have an existing Google Cloud Support package, then you already have support for Anthos and Anthos on bare metal.
For more information, see the Google Cloud Support documentation.
Requirements for Anthos on bare metal support
To troubleshoot business-critical incidents effectively:
- Check that your environment is current and within the published end-of-support timeframes. See the Version Support Policy section for more information.
- Enable Cloud Logging and Cloud Monitoring for system components. For details, see the following Support tools section.
To troubleshoot an Anthos on bare metal incident, Google Cloud Support relies on three pieces of information:
Your environment configuration
When you open a support case, running the following commands provides key information about your cluster setup:
For all your cluster types, run
bmctl check cluster --snapshotcommand to capture information about Kubernetes and your nodes. Attach the resulting tarball to the support case.
For admin, hybrid, and standalone clusters, run the
bmctl check clustercommand to check the health status of the cluster and nodes. Attach the resulting logs to the support case. They should exist under the
For user clusters, first create a health check YAML file with the cluster name and namespace, and then apply the file in the appropriate admin cluster:
- Create a YAML file with the following
healthcheckproperties. Here is sample content for a cluster named
apiVersion: baremetal.cluster.gke.io/v1 kind: HealthCheck metadata: generateName: healthcheck- namespace: cluster-user1 spec: clusterName: user1
- After you create the YAML file, apply the custom resource in the admin cluster
that is managing the user cluster with the
kubectlcommand. Here is a sample command using the YAML file created in the previous step. In the sample, the ADMIN_KUBECONFIG variable specifies the path to the admin cluster's kubeconfig file:
kubectl --kubeconfig ADMIN_KUBECONFIG create -f healthcheck-user1.yamlThe command returns the following response:
- Wait until the health check job is completed by testing to see if the
health check job has finished reconciling. In the previous example case,
the health check job name is
healthcheck.baremetal.cluster.gke.io/healthcheck-7c4qf. Here is a sample test with the
kubectlcommand that waits 30 minutes for the health check job to complete:
kubectl --kubeconfig ADMIN_KUBECONFIG wait healthcheck healthcheck-7c4qf -n cluster-user1 \ --for=condition=Reconciling=False --timeout=30mWhen completed, this command returns:
healthcheck.baremetal.cluster.gke.io/healthcheck-7c4qf condition metYou can see the health check job results with the following command:
kubectl --kubeconfig ADMIN_KUBECONFIG get healthcheck healthcheck-7c4qf -n cluster-user1The command returns the following result:
NAME PASS AGE healthcheck-7c4qf true 17m
- Gather all the health check job pods's logs into a local file with the
kubectlcommand. Here's an example using the previous sample health check job:
kubectl --kubeconfig ADMIN_KUBECONFIG logs -n cluster-user1 -l baremetal.cluster.gke.io/check-name=healthcheck-7c4qf --tail=-1 > healthcheck-7c4qf.log
- Create a YAML file with the following
When you create a new Anthos on bare metal cluster, Cloud Logging agents are enabled by default and scoped only to system-level components. This replicates system-level logs into the Google Cloud project associated with the cluster. System-level logs are from Kubernetes pods in the following namespaces:
kube-system gke-system gke-connect istio-system config-management-system gatekeeper-system cnrm-system knative-serving
Logs can be queried from the Cloud Logging console.
Note: If Cloud Logging is disabled, support is offered on a best-effort basis only, and could require significant additional effort from your on-site engineering team.
For more details, see Logging and Monitoring.
In addition to logs, metrics are also captured by the Cloud Monitoring agent. This replicates system-level metrics into the Google Cloud project associated with the cluster. System-level metrics are from Kubernetes pods running in the same namespaces listed in Logs.
Note: If Cloud Monitoring is disabled, support is offered on a best-effort basis only, and could require significant additional effort from your on-site engineering team.
For more details, see Logging and Monitoring.
How we troubleshoot your environment
Here is an example of a typical support incident:
- Someone—-for example, the cluster administrator—-opens a support case via
Google Cloud Console or the Google Cloud Support Center, and selects Anthos and
Anthos on bare metal as Category and Component, respectively. They enter the
information required and attach the output of relevant
bmctlcommands to the case.
- The support case is routed to a Technical Support Engineer specializing in Anthos on bare metal.
- The support engineer examines the contents of the snapshot to gain context of the environment.
- The support engineer examines the logs and metrics in the Google Cloud project, entering the support case ID as the business justification, which is logged internally.
- The support engineer responds to the case with an assessment and recommendation. The support engineer and the user continue troubleshooting until they come to a resolution.
What does Google support?
Generally, the Cloud Support team supports all software components shipped as part of Anthos on bare metal as well as Anthos Service Mesh and Anthos Config Management. The table below details this further:
|Google Cloud supported||Not supported|
|Kubernetes and the container runtime||Customer choice of load balancer (manual load balancing)|
|Connect and the Connect Agent||Customer code (see Developer Support below)|
|Google Cloud operations, Monitoring, Logging, and agents||Customer choice of operating system|
|Bundled load balancer||Physical or virtual server, storage, and network|
|Ingress controller||External DNS, DHCP, and identity systems|
|Anthos Identity Service|
|Anthos Service Mesh|
|Anthos Config Management|
Version Support Policy
To learn about the overall version support policy, see the Anthos support page
Shared Responsibility Model
Running a business-critical production application on Anthos on bare metal requires multiple parties to carry different responsponsibilities. While not an exhaustive list, the sections below list the roles and responsibilities.
- Maintenance and distribution of the Anthos on bare metal software package.
- Notifying users of available upgrades for Anthos on bare metal, and producing upgrade scripts for the previous version; Anthos on bare metal supports sequential upgrades only (example: 1.2 → 1.3 → 1.4 and not 1.2 → 1.4).
- Operating the Connect and Cloud Operations services.
- Troubleshooting, providing workarounds, and correcting the root cause of any issues related to Google-provided components
- Overall system administration for on-premises clusters.
- Maintaining any application workload deployed on the cluster.
- Running, maintaining, and patching the data center infrastructure, including networking, servers, operating system, storage, and connectivity to Google Cloud.
- Running, maintaining, and patching network load balancers if manual load balancer option is chosen.
- Upgrading Anthos on bare metal versions on a regular basis.
- Monitoring of the cluster and applications, and responding to any incidents.
- Ensuring Cloud Operations agents are deployed to clusters.
- Providing Google with environmental details for troubleshooting purposes.
Google does not provide support for application workloads running on Anthos on bare metal. However, we do provide best-effort developer support to ensure your developers can easily run applications on Anthos on bare metal. We believe that engaging earlier during development can prevent critical incidents later in the deployment.
This Developer Support is available to customers with a paid support package and is treated as a P3 priority for an issue blocking a launch, or a P4 priority for general consultation.