GitLab Technical Marketing, 3 Sep 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: ECS Task Definition from local JSON

Description

ECS Task Definition from local JSON
Introduced in GitLab 13.3, ECS Task Definition from local JSON allows you to have the ECS Task Definition file in your GitLab repository. This means that you can exploit our version control and collaboration capabilities to maintain this file. Likewise, you can extend your organizations audit and compliance regulations to the Task Definition of your ECS services, have many stakeholders collaborate on the configuration and development of the Task Definition, and rollback to a previous version of the file, if needed. Lastly, this new feature helps you streamline your AWS deployments.

A

Hello, my name is cesar cevedra, I'm a technical marketing manager here at gitlab. In this short presentation, I will give you an overview of the ecs task definition from local json capability introduced in gitlab 13.3, which was an improvement to our gitlab's ecs deployment template.

A

This new capability introduces the ability to update the task definition of an ecs service with a json file defined in your local repository, but before we jump into this new capability introducing 13.3, let's quickly revisit how you set up an ecs cluster and also how do you integrate it with gitlab?

A

So, if you check out our documentation for the integration, our integration with ecs, you will notice that you actually need to define three variables in your project: ecs cluster service and test definition, and in my case I have this project.

A

And here I have ecs cluster.

A

Which is called sisabedra acs cluster that I already defined and I'm sorry created.

A

Then you have ecs service in this demo. I'm going to be talking about the an ecs service that is for fargate and that's the environment of this variable is going to be review, wildcard environments and then the task definition which is called sees of address spring task, def f, so those are the three variables needed.

A

Obviously, beyond that, you also need the the aws access key region and secret access key to be able to connect to aws uh from from the runner. One more thing you need is the auto devops platform target. I have auto devops enabled here I'm because I'm using the ecs deployment template and in this case I've defined the value to be fargate, because I'm going to be creating a fargate target test definition and a farget fargate ecs service.

A

I have duplicates here because they are targeted to different environments. As you can see. The next question is: how do you create? uh You know? You have to create ecs service, the cluster first uh service test definition.

A

So uh one way to do it is uh you know, following the um aws documentation, and the first thing that uh personally have I create is uh the task definition and the test definition is a description of what's going to be in in a container that is going to be running as a service in acs.

A

So you follow the instructions uh they actually have uh launch wiz uh wizards in aws, so you can just go through each page and define uh the values as, as you know, per the requirements of your application, then the next thing I did was create the cluster and again there are instructions on how to clear the cluster, and I chose ec2, linux plus networking, and then, after that, I created an application load balancer that would take in traffic from the internet and route it to a specific ecs service.

A

In this case I have an ecs service per environment, so I have three load balancers and then finally, I created the serv, the ecs service, and they actually, you can follow the documentation as well as the wizards that the aws provides.

A

So, let's just let me show you what it looks like that looks like once they are created. So here is this see surveyor ics cluster? This has three services. I have three environments, so my production environment is actually deploying my application to this running service. I have another branch for review, which is this one here.

A

The review that starts with a two roll back that environment is being deployed to this right here to this service and the far gate based service is being deployed to this uh instance. Here, the first iteration of of our integration to ecs was the ability to deploy from gitlab to ecs through the pipeline, and it looked kind of like this.

A

So here so so, this is a pipeline, a deployment pipeline and it's building this application here that happens to be a java spring sample application, and then it deploys it to fargate to uh it's basically deploying it to this service.

A

Okay and it's using the task definition name c sub editor screen task def. So these are the parameters that are being defined in the environment, variables of the project to the cluster, the service name and the test definition. The improvement introduced in 13.3 to this integration is the ability to actually have the content of the task definition, which is right here.

A

Let's open the one from fargate, for example, you can pick the last one here to have this file. This json file define and actually declared within your project.

A

So we're going to do in this.

A

In this demo is we're going to be creating a file and and the file content will be the task definition and then, when we do the deployment to ecs, the deployment will use that file that you created here in your project to push out the deployment to ecs instead of just reusing the one that was already here.

A

A

And you know this allows you to manage your definition file test definition file uh within your project. You know you get the versioning. Obviously you can you can um you know, track the changes. um You have call stakeholders collaborate in mrs uh to modify and update the test definition file and also the streams lines, streamlines your deployment to ecs, making uh you more productive as a developer. So the next thing I'm going to show you is how the pipeline looks like so the pipeline here is super simple.

A

It just includes a template: okay, which is basically the gitlab ccs deployment, template, there's a bunch of variables here to disable uh a lot of the tests that are included in the autodevops pipeline. So if we look at the um the ec deploy ecs git lab here, this contains uh five stages and it includes uh actually it is setting up the devops platform to ecs, which is not going to be used.

A

I'm using remember I'm using fargate in the project, so that's going to be overwritten, and then it includes these two templates and the one of interest here. Is this one here and this template is actually the one doing the deployment uh there are different types of deployments. It could be uh just a production, ecs production, fire gate, review, fargo, review, ecs, there's those four options and what it's really doing is doing an ecs update test definition. So it's going to think of it.

A

It's going to be replacing the running service by applying the new task definition file from your gitlab project. So, let's start making the changes then.

A

So, in this case, let's add the task definition file to our project. So first, let's create a directory and then inside that directory. Let's click create a new file.

A

So how did I go about creating this task definition file?

A

Well, one way to do it is to actually go to the aws documentation for ecs, test definition, creation and copy the template and then adjust the template according to your application needs and then use that as the content of the file that you will be storing in your repository, which is this one here, but what I did was, which is another way, is to use the amazon ecs test definition console you create on click on create test definition, you select fire gate, and then you follow the wizard entering the the name of your test definition, the requirements of your application and container.

A

Then, at the end, you can click on this button and it'll give you the json file for uh all everything that you selected in the through the wizard. You can copy and paste that json onto here now. What I did was I copy that json, but that json needs to be mod slightly modified.

A

This is the original json that I copy and pasted, and one thing I had to do is delete every single line that had a value equal to null, so this is the before and after so notice, this file has no null values.

A

Another thing I did was: I had to delete the section that says repository credential, so I deleted this, so it wouldn't get in the way of the runner correctly signing in to the awscs environment, and the last thing I did was I had to make sure that the aws log groups had the um the location of the task definition that we were replacing okay, so this is the think of this is the before and then the new we're replacing that test definition with this new new one.

A

Something to keep in mind here, though, is that you need to keep the log location the way it was for this to work.

A

All right, so, the next thing that we need to do here is actually define a variable and the new environment.

A

Variable needs to be ci, aws, ecs task definition.

A

A

And then the value is going to be ci and then the file name is td, ecs, fargate, dot, json, oh and then we want this to.

A

A

There ddes fargate json right now. I have a running service under one of my.

A

Environments under this environment, which is the one I want to affect with this change it says spring, is here in blue okay background is blue, so just make let's make a change for the of the background.

A

And one more thing is that if we go to.

A

The cluster right now this uh this is the service name that I'm trying to redeploy. I will be redeploying, and the task definition is c subvedra spring task: def, uh f, okay for fargate.

A

So after the deployment, this test definition name will change, because the new name will.

A

Be will be uh this one right here, so the new name will be sisa. Vedra fargate task, def f, not spring notice that this is a new name. Okay, and I'm showing you this so that you can see the difference once the deployment happens.

A

So let's make a change to the code so that we can kick off a pipeline.

A

We need to go here, let's just go to web ide and let's make a change.

A

We'll make this we'll turn this to brown.

A

Okay and then we also need to change it.

A

A

All right, so then we commit the changes to the same branch.

A

And this is going to kick off a pipeline.

A

Which is going to build and it will deploy to fargate now that the pipeline is done executing let's go into the review, fargate.

A

Job and notice that it succeeded.

A

The task definitions that has been updated with the file from our project.

A

And let's check on the amazon ecs and here, as you can see,.

A

The ecs service dash f here as before no longer has this c sub editor spring test definition f. Now it has the c sub, far gate task, def dash, f and version number one, because this is the first version of this test definition that has been applied to this service. If we look at the tasks notice that the new task is on its way up, so we need to wait until it's in a running state for us to see the new newly deployed web page.

A

Now that the uh fargate server task is up and running, we can go and check the web application.

A

To ensure that the changes that we applied earlier have been deployed and are successfully up and running, and there it is. This is the updated background, so the changes have been deployed using the new task definition file by the way, just as you you can use and declare a file a json file here.

A

Another way to provide the json file for the test definition is to do it as a variable.

A

So in that case you would come to the variable here variable section, and here you can actually paste the content of the json file here and make sure that you change this to file and that should also work.

A

So, in conclusion, in this technical demo, we have discussed the new feature, ecs task definition from local json and having the task definition file on your gitlab repository, allows you to maintain its versions via our version. Control and collaboration capabilities have other stakeholders, collaborate on the task, definition configuration and development roll back to a previous version of a task definition and can also help you with your organizational audit and compliance of your ecs container definitions.

A

I hope you enjoyed this video and I'll, see you next time. Thank you.