[MUSIC PLAYING]
[MUSIC PLAYING]

SPEAKER: In this video, you will learn about Kubernetes events,
SPEAKER: In this video, you will learn about Kubernetes events,

their use cases, how to view them,
their use cases, how to view them,

and how to use them to troubleshoot
and how to use them to troubleshoot

a common issue with Google Kubernetes Engine,
a common issue with Google Kubernetes Engine,

or GKE, clusters.
or GKE, clusters.

Let's get started by understanding
Let's get started by understanding

what Kubernetes events are.
what Kubernetes events are.

Kubernetes events are objects that
Kubernetes events are objects that

give insights about what's happening inside a GKE cluster.
give insights about what's happening inside a GKE cluster.

They are usually generated when a cluster's resources,
They are usually generated when a cluster's resources,

such as nodes, Pods, or containers, change state.
such as nodes, Pods, or containers, change state.

These events can be used to monitor your application
These events can be used to monitor your application

and cluster state, respond to failures,
and cluster state, respond to failures,

and perform diagnostics.
and perform diagnostics.

Remember, events have a limited retention time
Remember, events have a limited retention time

and will eventually be deleted.
and will eventually be deleted.

So you must monitor and collect events as they happen.
So you must monitor and collect events as they happen.

Kubernetes events can be useful in multiple scenarios.
Kubernetes events can be useful in multiple scenarios.

Here are a couple common use cases.
Here are a couple common use cases.

Pods not running within your cluster--
Pods not running within your cluster--

viewing the Kubernetes events for a Pod
viewing the Kubernetes events for a Pod

can reveal why it is not running,
can reveal why it is not running,

leading to a quick resolution.
leading to a quick resolution.

For example, you may see ImagePullBackOff,
For example, you may see ImagePullBackOff,

indicating the cluster or node is
indicating the cluster or node is

unable to access the repository where the container image is
unable to access the repository where the container image is

hosted.
hosted.

A node in your cluster with a Not Ready status--
A node in your cluster with a Not Ready status--

viewing the Kubernetes events can indicate
viewing the Kubernetes events can indicate

why the node is not ready.
why the node is not ready.

For example, a node might have disk pressure,
For example, a node might have disk pressure,

indicating that the underlying node is running out
indicating that the underlying node is running out

of disk space.
of disk space.

Kubernetes events can be viewed in three different ways
Kubernetes events can be viewed in three different ways

with a GKE cluster--
with a GKE cluster--

in the Kubernetes Engine Console,
in the Kubernetes Engine Console,

Logging Console, and within the Terminal using Kubectl.
Logging Console, and within the Terminal using Kubectl.

Before you can view the Kubernetes events,
Before you can view the Kubernetes events,

ensure that users are granted with the appropriate identity
ensure that users are granted with the appropriate identity

and access management or IAM roles.
and access management or IAM roles.

Viewing events using the Kubernetes Engine
Viewing events using the Kubernetes Engine

Console requires the following IAM roles--
Console requires the following IAM roles--

Kubernetes Engine Viewer, Kubernetes Engine Cluster
Kubernetes Engine Viewer, Kubernetes Engine Cluster

Viewer, and Logging Logs Viewer.
Viewer, and Logging Logs Viewer.

Viewing events using the Logging Console requires
Viewing events using the Logging Console requires

only the following IAM roles--
only the following IAM roles--

Logging Logs Viewer.
Logging Logs Viewer.

To view Kubernetes events in the Terminal using Kubectl,
To view Kubernetes events in the Terminal using Kubectl,

your credentials must have at least the default cluster role
your credentials must have at least the default cluster role

view or a cluster role that allows access
view or a cluster role that allows access

to the Resource Events and the ability
to the Resource Events and the ability

to call get, list, and watch.
to call get, list, and watch.

Let's now look at how to view Kubernetes events.
Let's now look at how to view Kubernetes events.

Within Google Cloud, you can view events
Within Google Cloud, you can view events

across all levels of the Kubernetes hierarchy,
across all levels of the Kubernetes hierarchy,

including cluster, nodes, and workloads.
including cluster, nodes, and workloads.

Remember, for Kubernetes clusters,
Remember, for Kubernetes clusters,

events are only available for about 60 minutes.
events are only available for about 60 minutes.

For GKE, events are exported and are available using the Logging
For GKE, events are exported and are available using the Logging

Console.
Console.

Next, you can view Kubernetes events
Next, you can view Kubernetes events

in Cloud Logging with the following query--
in Cloud Logging with the following query--

log_ID("events") and resource.label.cluster
log_ID("events") and resource.label.cluster

name="CLUSTER NAME".
name="CLUSTER NAME".

Finally, you can view Kubernetes events from Kubectl
Finally, you can view Kubernetes events from Kubectl

by running kubectl events -A or, for a specific resource,
by running kubectl events -A or, for a specific resource,

kubectl events --for the resource
kubectl events --for the resource

forward slash the resource name.
forward slash the resource name.

For example, kubectl events --for pod/ubuntu.
For example, kubectl events --for pod/ubuntu.

Another useful way to view events
Another useful way to view events

is with the describe command for a specific resource.
is with the describe command for a specific resource.

For example, kubectl describe pod ubuntu.
For example, kubectl describe pod ubuntu.

The Kubernetes events for the specific resource
The Kubernetes events for the specific resource

are included at the bottom of the output.
are included at the bottom of the output.

Let's dive into a demo and check out
Let's dive into a demo and check out

how to use Kubernetes events for troubleshooting a common issue
how to use Kubernetes events for troubleshooting a common issue

in the GKE cluster.
in the GKE cluster.

This cluster has a number of workloads running,
This cluster has a number of workloads running,

but there appears to be one workload with no Pods running.
but there appears to be one workload with no Pods running.

Let's get the Kubernetes events for the Pod in this deployment
Let's get the Kubernetes events for the Pod in this deployment

by running the following command--
by running the following command--

kubectl events -n k8sevents --for the Pod, forward slash,
kubectl events -n k8sevents --for the Pod, forward slash,

and the Pod name.
and the Pod name.

Oh, it looks like there's insufficient memory
Oh, it looks like there's insufficient memory

in the cluster for the Pod to be scheduled.
in the cluster for the Pod to be scheduled.

The solution here is to scale up.
The solution here is to scale up.

So let's add more nodes.
So let's add more nodes.

Now that the additional notes are ready,
Now that the additional notes are ready,

let's check the status of the workload.
let's check the status of the workload.

Hm, eventsdeployment workload still has no Pods running.
Hm, eventsdeployment workload still has no Pods running.

Let's view the events again.
Let's view the events again.

The pods are unscheduled, as they could not
The pods are unscheduled, as they could not

meet the taint and toleration requirement to be
meet the taint and toleration requirement to be

scheduled on the new node.
scheduled on the new node.

Let's add the toleration to the workload.
Let's add the toleration to the workload.

Success.
Success.

All the Pods in the workload are running.
All the Pods in the workload are running.

This is just one scenario where Kubernetes events assisted us
This is just one scenario where Kubernetes events assisted us

in quickly identifying and resolving
in quickly identifying and resolving

our issue of a deployment not running.
our issue of a deployment not running.

The next time you see a problem with your cluster,
The next time you see a problem with your cluster,

try using Kubernetes events to identify and resolve issues.
try using Kubernetes events to identify and resolve issues.

For more details about Kubernetes events,
For more details about Kubernetes events,

check out the following documentation.
check out the following documentation.

Thanks for watching.
Thanks for watching.

[MUSIC PLAYING]
[MUSIC PLAYING]