1

00:00:01,050 --> 00:00:04,140

Hello and welcome to this lecture. In this lecture



2

00:00:04,170 --> 00:00:10,370

we talk about Monitoring a Kubernetes cluster.



3

00:00:10,410 --> 00:00:16,890

So how do you monitor resource consumption on Kubernetes? Or more importantly what would you like



4

00:00:16,890 --> 00:00:18,300

to monitor?



5

00:00:18,300 --> 00:00:23,330

I’d like to know Node level metrics such as the number of nodes in the cluster,



6

00:00:23,460 --> 00:00:31,500

how many of them are healthy as well as performance metrics such as CPU. Memory, network and disk utilization.



7

00:00:32,160 --> 00:00:38,670

As well as POD level metrics such as the number of PODs, and performance metrics of each POD such



8

00:00:38,670 --> 00:00:41,940

as the CPU and Memory consumption on them.



9

00:00:41,940 --> 00:00:48,600

So we need a solution that will monitor these metrics store them and provide analytics around this data



10

00:00:49,830 --> 00:00:56,040

As of this recording, Kubernetes does not come with a full featured built-in monitoring solution.



11

00:00:56,040 --> 00:01:03,510

However, there are a number of open-source solutions available today, such as the Metrics-Server, Prometheus,



12

00:01:04,020 --> 00:01:09,450

Elastic Stack, and proprietary solutions like Datadog and Dynatrace.



13

00:01:12,580 --> 00:01:18,870

Heapster was one of the original projects that enabled monitoring and analysis features for kubernetes



14

00:01:18,870 --> 00:01:25,640

You will see a lot of reference online when you look for reference architectures on monitoring Kubernetes.



15

00:01:25,780 --> 00:01:33,580

However, Heapster is now Deprecated and a slimmed down version was formed known as the Metrics Server.



16

00:01:33,580 --> 00:01:40,310

You can have one metrics server per kubernetes  cluster the metric server retrieves metrics from each



17

00:01:40,310 --> 00:01:46,100

of the kubernetes nodes and pods, aggregates them and stores them in memory.



18

00:01:46,100 --> 00:01:52,340

Note that the metric server is only an in memory monitoring solution and does not store the metrics



19

00:01:52,400 --> 00:01:57,480

on the desk and as a result you cannot see historical performance data.



20

00:01:57,650 --> 00:02:04,760

For that you must rely on one of the advanced monitoring solutions we talked about earlier in this lecture.



21

00:02:04,760 --> 00:02:10,050

So how are the metrics generated for the PODs on these nodes?  Kubernetes



22

00:02:10,060 --> 00:02:17,040

runs an agent on each node known as the kubelet, which is responsible for receiving instructions



23

00:02:17,040 --> 00:02:22,430

from the kubernetes API master server and running PODs on the nodes.



24

00:02:22,650 --> 00:02:30,750

The kubelet also contains a subcomponent known as as cAdvisor or Container Advisor.  cAdvisor is



25

00:02:30,750 --> 00:02:37,140

responsible for retrieving performance metrics from pods, and exposing them through the kubelet API



26

00:02:37,200 --> 00:02:44,480

to make the metrics available for the Metrics Server. If you are using minikube for your local cluster,



27

00:02:44,840 --> 00:02:46,850

run the command minikube



28

00:02:47,000 --> 00:02:54,770

addons enable metrics-server. For all other environments deploy the metrics server by cloning the



29

00:02:54,770 --> 00:03:00,950

metrics-server deployment files from the github repository. And then deploying the required components



30

00:03:01,310 --> 00:03:08,810

using the kubectl create command.  This command deploys a set of pods,  services and roles to enable



31

00:03:08,810 --> 00:03:14,150

metrics server to poll for performance metrics from the nodes in the cluster.



32

00:03:14,150 --> 00:03:22,190

Once deployed, give the metrics-server some time to collect and process data. Once processed, cluster performance



33

00:03:22,190 --> 00:03:27,700

can be viewed by running the command kubectl top node.



34

00:03:27,770 --> 00:03:34,850

This provides the CPU and Memory consumption of each of the nodes. As you can see 8% of the CPU



35

00:03:34,850 --> 00:03:40,480

on my master node is consumed, which is about 166 milli cores.



36

00:03:40,820 --> 00:03:48,730

Use the kubectl top pod command to view performance metrics of pods in kubernetes.



37

00:03:48,890 --> 00:03:50,420

That's it for this lecture.



38

00:03:50,420 --> 00:03:55,740

Head over to the coding exercises section and practice viewing performance metrics on the kubernetes



39

00:03:55,740 --> 00:03:57,120

cluster.



40

00:03:57,170 --> 00:03:57,660

Thank you.