You are viewing documentation for Kubeflow 0.5

This is a static snapshot from the time of the Kubeflow 0.5 release.
For up-to-date information, see the latest version.

Logging and monitoring

Logging and monitoring for Kubeflow

This guide has information about how to set up logging and monitoring for your Kubeflow deployment.

Logging

Stackdriver on GKE

The default on GKE is to send logs to Stackdriver logging.

Stackdriver recently introduced new features for Kubernetes Monitoring that are currently in Beta. These features are only available on Kubernetes v1.10 or later and must be explicitly installed. Below are instructions for both versions of Stackdriver Kubernetes.

Default stackdriver

This section contains instructions for using the existing stackdriver support for GKE which is the default.

To get the logs for a particular pod you can use the following advanced filter in Stackdriver logging’s search UI.

resource.type="container"
resource.labels.cluster_name="${CLUSTER}"
resource.labels.pod_id="${POD_NAME}"

where ${POD_NAME} is the name of the pod and ${CLUSTER} is the name of your cluster.

The equivalent gcloud command would be

gcloud --project=${PROJECT} logging read  \
     --freshness=24h \
     --order asc \
        "resource.type=\"container\" resource.labels.cluster_name=\"${CLUSTER}\" resource.labels.pod_id=\"${POD}\" "

Kubernetes events for the TFJob are also available in stackdriver and can be obtained using the following query in the UI

resource.labels.cluster_name="${CLUSTER}"
logName="projects/${PROJECT}/logs/events" 
jsonPayload.involvedObject.name="${TFJOB}"

The equivalent gcloud command is

gcloud --project=${PROJECT} logging read  \
     --freshness=24h \
     --order asc \
        "resource.labels.cluster_name=\"${CLUSTER}\" jsonPayload.involvedObject.name=\"${TFJOB}\" logName=\"projects/${PROJECT}/logs/events\" "

Stackdriver Kubernetes

This section contains the relevant stackdriver queries and gloud commands if you are using the new Stackdriver Kubernetes Monitoring

To get the stdout/stderr logs for a particular container you can use the following advanced filter in Stackdriver logging’s search UI.

resource.type="k8s_container"
resource.labels.cluster_name="${CLUSTER}"
resource.labels.pod_name="${POD_NAME}"

where ${POD_NAME} is the name of the pod and ${CLUSTER} is the name of your cluster.

The equivalent gcloud command would be

gcloud --project=${PROJECT} logging read  \
     --freshness=24h \
     --order asc \
        "resource.type=\"k8s_container\" resource.labels.cluster_name=\"${CLUSTER}\" resource.labels.pod_name=\"${POD_NAME}\" "

Events about individual pods can be obtained with the following query

resource.type="k8s_pod"
resource.labels.cluster_name="${CLUSTER}"
resource.labels.pod_name="${POD_NAME}"

or via gcloud

gcloud --project=${PROJECT} logging read  \
     --freshness=24h \
     --order asc \
        "resource.type=\"k8s_pod\" resource.labels.cluster_name=\"${CLUSTER}\" resource.labels.pod_name=\"${POD_NAME}\" "

Filter with labels

The new agents also support querying for logs using pod labels For example:

resource.type="k8s_container"
resource.labels.cluster_name="${CLUSTER}"
metadata.userLables.${LABEL_KEY}="${LABEL_VALUE}"

Monitoring

Stackdriver on GKE

The new Stackdriver Kubernetes Monitoring provides single dashboard observability and is compatible with Prometheus data model.

See this doc for more details on the dashboard.

Stackdriver by default provides container level CPU/memory metrics. We can also define custom Prometheus metrics and view them on the Stackdriver dashboard. See for more detail.

Prometheus

Kubeflow Prometheus component

Kubeflow provides a Prometheus component. To deploy the Prometheus component:

ks generate prometheus prom --projectId=YOUR_PROJECT --clusterName=YOUR_CLUSTER --zone=ZONE
ks apply YOUR_ENV -c prom

The prometheus server will scrape the services with annotation prometheus.io/scrape=true. See for more detail and an example.

Export metrics to Stackdriver

The Prometheus server will export metrics to Stackdriver, as configured. We are using an image provided by Stackdriver. See Stackdriver doc for more detail, but you don’t need to change anything here.

If you don’t want to export metrics to Stackdriver, remove the remote_write part in the prometheus.yml, and use a native Prometheus image.

Metric collector component for IAP (GKE only)

Kubeflow also provides a metric-collector component. This component periodically pings your Kubeflow endpoint and provides a metric of whether the endpoint is up or not. To deploy it:

ks generate metric-collector mc --targetUrl=YOUR_KF_ENDPOINT
ks apply YOUR_ENV -c mc