Cloud Native Computing Foundation PrometheusDay EU 2022, 19 May 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Lightning Talk: Monitoring Counter Strike Global offensive with Prometheus - David Lorite, Sysdig

Description

Lightning Talk: Monitoring Counter Strike Global offensive with Prometheus - David Lorite, Sysdig

Everyone is using Prometheus in their infrastructure, but who is using Prometheus in their game server? In gaming, servers are a critical component of the industry's success. The gaming industry is highly profitable and the enabling technology is critical to its success. They also carry a great responsibility in maintaining quality of service (QoS), where a drop in the latency or in the computing power, especially in multiplayer games, seriously affects user experience. can be critical. In this talk, you will learn how to set up and monitor a Counter-Strike: Global Offensive server with Prometheus. We will show the installation and configuration of the Prometheus server and the following exporters: - Node exporter: to monitor the infrastructure metrics. - CAdvisor: to monitor the usage of the containers. - SRCDS Exporter: to monitor the game server metrics. With all these exporters, apart from monitoring the game itself, we will have visibility into the node and the applications on it, to be sure that the VM is running everything at an optimal service level and avoids extra costs in our cloud bill.

A

Thank you, everyone for joining this little intel today, I'm david lorite and I work as an interrogation, engineering and system.

A

uh I want to to show you how I'm I monitor a cs. Go server with prometheus. We describe it. We wanted to. We started this spread because we wanted to describe how to monitor a full stack of monitoring with prometheus.

A

A

We will have fun collecting these metrics. We learn more about how promises works.

A

People there are people when they start learning about how prometheus exporter works. They are confusing because they don't understand very well what are the sporters and hogwarts.

A

So we want to describe that and even that we might save some money, because sometimes you create an instance and you you suddenly realize that the instant you created before was very huge and you can recite it.

A

So we prepared we wanted when you want to know if your application is performing well, you probably could see if your note held it's okay and then the last step, maybe you could check, is your container have no resources?

A

Last you can focus on your application.

A

So we prepared a docker compose with five containers that we were deployed with the project, and these containers are the game server, that is, the cs, go server, the promised server and the exporters.

A

In this case, for the exporters we use it for the node, we use it. The node exporter for the containers we use at the csvisor and for the application that, in this case, is the cs. Go is the cr3ds exporter the.

A

If we are focusing the crts exporter, we saw that suppose a few metrics so which of those metrics we want to monitor. We will start checking the number of people that is playing the game at the moment.

A

We are also could check the map that is being played at the moment. Also, and also the the metric in this case is always always half the the value one you can check in the label.

A

The name of the of the map is in play and if we are focusing the container metrics and do when you are focused in the in the container magnetrix, you want to assure that the container has enough resources.

A

For example, if you see how the memory is doing, you can see when the starts begin, the memory is increasing and the cpu is more straightforward.

A

The cpu in the beginning, it's either and when a player joined to the game, the cpu increase, and when, when you see the graph you can see when the game starts and when ends also, you can see which map is consuming more cpu than the others. That's checking the the metric in the case of the nodes metrics it's more difficult to see, for example, for the memory it's difficult to see anything because the other containers in the host are consuming more memory than the game server.

A

On the other hand, for the cpu, it's more clear you can see when the game starts when it ends.

A

Okay and a common mistake that we do when you don't know how many resources your application is using, it's creating an instance with more the result than you need. In order to take care of this, you can check the overview of the resources you are using and recite it. Accordingly.

A

I told you that you might save some money and you can so.

A

To let's sum up, we deployed a game server in a cloud instance, we get the metrics for uh with with our prometheus. We learn which metrics are important, and this approach was so fun.

A

I made a the next time I may deploy it in a kubernetes cluster, so I can may be a true glove native gamer, but that's for the next time, and that's all thanks. Everyone for listening have a happy evening.