Collecting Anthos Service Mesh logs

The following sections explains how to gather the various Anthos Service Mesh logs for troubleshooting issues or contacting Google Support.

Collect logs using the bug report tool

Anthos Service Mesh provides an automated bug report tool that collects the relevant diagnostic logs and lets you attach the logs to a Google Support ticket.

Before you begin, ensure your kubeconfig context is set to the target cluster.

Verify your context using the following command:

kubectl config current-context

Start the log collection

To start the log collection, run the bug-report tool using the following command:

istioctl bug-report

Upload your debug archive

The tool creates an archive of your mesh's logs and configuration in the working directory. You can unpack the archive and use the troubleshooting guides to attempt to perform troubleshooting yourself. However, if you have a support package, you can contact Google Cloud Support, who will provide you with further steps to securely upload your log archive.

Manually collect Anthos Service Mesh logs

Instead of using the Anthos Service Mesh bug report tool, this section explains how to manually collect all the relevant logs.

Envoy access logs

Envoy proxy access logs contain detailed information that is useful for troubleshooting. However, you must enable them and set the correct detail level.

For details about how to interpret the log contents, see Interpret Envoy logs.

Enable or disable Envoy logs

To enable the Envoy proxy access logs, configure an overlay file for in-cluster Anthos Service Mesh or a ConfigMap for managed Anthos Service Mesh.

Increase logging detail

To temporarily increase the detail level of the logs, use the following command. This setting is undone when the pod is recreated.

kubectl -n NAMESPACE exec POD_NAME -c istio-proxy -- curl -X POST http://localhost:15000/logging?level=debug

To set the detail level of the logs back to default, use the following command:

kubectl -n NAMESPACE exec POD_NAME -c istio-proxy -- curl -X POST http://localhost:15000/logging?level=info

Write Envoy logs to a folder

To collect the Envoy proxy access logs and store them in a folder, use the following command:

kubectl logs -l app=APPLICATION_NAME -c istio-proxy > /FILE_PATH

See Getting Envoy's Access Logs for more information.

Kubernetes logs

Kubernetes generates several logs that contain information about the behavior of Istio components, such as istiod, Ingress Gateway, and proxies. You can review these logs for errors, which might narrow the scope of possible causes of a problem.

Capture istiod logs using the following command:

kubectl -n istio-system logs $(kubectl -n istio-system get pods -lapp=istiod -oname) > ./LOGS_FOLDER/istiod.log

Capture Istio Ingress Gateway logs using the following command:

kubectl -n istio-system logs $(kubectl -n istio-system get pods -lapp=istio-ingressgateway -oname) > /FILE_PATH

Capture Istio Proxy logs using the following command:

kubectl -n WORKLOAD_NAMESPACE logs POD_NAME -c istio-proxy > ./LOGS_FOLDER/proxy.log

Kubernetes configuration dump

This information allows users without direct access to the cluster to view the state of various resources and identify possible configuration problems. The following command writes the Kubernetes configuration to a YAML file:

for ns in `kubectl get namespaces -o=jsonpath='{.items[*]}'` ; do echo "===NAMESPACE===" $ns >> ./LOGS_FOLDER/kubernetes.yaml ;kubectl get -oyaml -n $ns deploy,statefulset,job,ingress,endpoints,configmap,event,secret,service,istio-io >> ./LOGS_FOLDER/kubernetes.yaml; done

Envoy core dump

Envoy core dumps are not typically useful to end users, however Google Support might request that you collect it as part of the troubleshooting process, using the following steps.

To configure the kernel to write Envoy core dumps to a writable directory:

  1. Add the label to a pod.

  2. Restart the pod to enable Envoy core dumps.

  3. Copy the core dump out of the pod.

Configure Envoy proxy

The detailed Envoy proxy configuration contains additional detail that might be helpful for troubleshooting purposes. You can collect this information using the following command. In this example, ENDPOINT is one of the following (shown in order of importance):

  • /certs
  • /clusters
  • /listeners
  • /config_dump
  • /memory
  • /server_info
  • /stats/prometheus
  • /runtime
kubectl exec -i POD_NAME -c istio-proxy -n NAMESPACE curl > out.log