GitLab Distribution Team Demos, 27 Jul 2023

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: Distribution Team Demo - July 27 - Testing new features as a developer

Description

As a developer, how to enable and test new feature (using container registry as an example, demonstrate setting up the metadata database https://gitlab.com/gitlab-org/charts/gitlab/-/blob/master/doc/charts/registry/index.md#database

A

Hello, everyone welcome to the distribution team demo on 27th July 2023, so this uh demo is also part of the series that we have been doing regarding distribution workflows, and this one will be focusing on how to enable a new feature based on the documentation and how to test it using our hand charts.

A

So this one is a follow-up to the previous video that that Mitch did, which showed how to spin apegit lab instance on gke using our hand charts. So we'll be continuing from that point where we have a functional gitlab instance, which runs gitlab. The latest version of gitlab using hand charts, so the demonstration uses a feature that is not enabled by default out of the box, so we'll be looking at enabling the metadata database for container page speed.

A

So this is an experimental feature that we added recently and we will be showing how to how to enable that, in our instance and how to make sure that it is working. So let me share my screen.

A

Okay, so this is our official documentation, the charts documentation for the container registrates out and in the that's, a subsection called database which uh describes how to how to enable the method at the database for container registry.

A

So, if we look at it, we can see it has a sample values.tml entries that should go until the registry key so yeah. It shows how to enable database set the host port, a user, how to specify password the SSL connection, credentials and configuration and all those things uh in addition to passing these values to the deployment, because it is an experimental feature and node fully polished. It requires some manual actions, so the manual actions being one we need to manually create a database for the registrator store metadata.

A

Do we need to create the user for to for the registry to access this database so and the credentials of this user will be stored as a kubernetes secret, so it the documentation, also has instructions regarding that how to create equivalent to secret with the password how to fall into our post Crystal, to create a database, create a user specify your ownership of the database and all those things so we'll be going through these steps and showing how the metadata database is functioning at the end of the demo, so uh switching to my terminal.

A

So this is a similar to what we were. What we did in the previous video, where we deployed a gitlab instance uh using gke with our hand, charts so on the bottom left corner. This is the configuration that I used here, which is very similar to what was done in the previous one previous video, and we can see in kns all the ports are running fine and without any issues.

A

So uh the manual steps, if you look at the documentation, the manual steps start with creating a secret with the database password, so it uses the cube. Control creates secret. This is a generic secret, not a dedicated one like a TLO certificate or anything uh a secret name from literal, the key in the secret, where the password will be stored and a password string.

A

So, let's, let's try doing that.

A

So yeah Cube control because we deployed it to the gitlab namespace. We will be passing the namespace, the create secret generic. This is the name of the secret that I am using from little, and this is the key I'm using password as the key and the value for the password I used a simple one called too many secrets.

A

Secure one, if you are doing this in production,.

A

So yeah, let's confirm that it got created. If we look at.

A

Okay, there is the gitlab instant database password.

A

Let's try decoding it and yeah it has the password key and too many secrets. That's the password value which matches what we did. The next one is login to your database instance. uh The command here is a two-part one. uh The inner one identifies the port name of your postgresql port and the outer one folds into the shell in that port. So if we look at the reports, we can say that gitlab possible scale. 0 is the port name. So, let's.

A

Fall into that port.

A

Okay, we have, we are inside the positive sport. Now we need a psql, a shell, to create the database and create the user again in the documentation. We have specified how to get a pay scale. Shell, which we copy- and here we have for we have fallen into the psql share.

A

So let's create a database user I am giving the username, as the registry itself created the role we need to set the password. The documentation specifies to paste the password in case you forgot what you gave, or in case you don't have it, but since we know what ours was, we can do it.

A

So I am giving the same too many secrets. That's password.

A

Now, let's create the database and specify the user registry at the order, so both the database name and the username is so you can give whatever you want so yeah. We created the database, let's come out of for the psql and the port, so yeah we have completed all the manual steps required. Now we need to specify this configuration values according to our need. uh The documentation shows all the available configurations for demo purposes. I am using a minimal version of that.

A

uh So I added the following to my Value Store PML file. uh The registry K database enabled because this is demo I am not using any SSL uh connection between registry and the database or postgresql the head disabled SSL mode, uh the user. So this is the database user that we created earlier with username registry and password. The password is in the secret that we created with the password key being what we specified password.

A

So, let's save that configuration file and we need to deploy the charts again.

A

So this is the same command. We use to install charge the first day because we use Helm upgrade dash dash, install uh it serves both purposes if it is a reference written, it installs the chart, if it is an upgrade, it detects that and updates the deployment with the new configuration values.

A

So, let's, let's try upgrading I will enter with our deployment with this new configuration now this this can be a bit time consuming, depending on various factors like uh network connectivity or the load on the cluster and all those things, but hopefully uh in the ports that are running in running multi-cluster. We should be seeing newer, newer ports coming up related to the metadata database and the old ones being deleted.

A

uh Hell is intelligent enough to detect if there are any changes to any other services. If there are none, those spots will be left as is, and only the digestible related ones will be deleted and new ones deployed.

A

So hopefully the two latest reports that we see in this cluster will get deleted and the new ones will be deployed. Let's now, we wait.

A

Because we can see newer jobs coming up, one is an upgrade check, because this is technically an upgrade from a previous installation. We enforce certain upgrade paths, so we have a job.

A

Upgrading from the supported minimum version to the current world.

A

So we can say changes happening in the cluster, so they help update command itself has not accepted successfully, so it is still going on.

A

Okay, we can see new registry ports coming up.

A

Let's say: registering migrations one: let's, let's check the logs on that one. We can see migrations happening on the metadata database.

A

And we can see new, newer ports coming up without any issues.

A

And the old ones like once the newer ports are up and running and we satisfy the requirements of the digits reports. The older ones will get automatically terminated like we see here. The older one is getting terminated.

A

And the next one is spinning up.

A

The other one is also terminating so yeah. uh We have successfully deployed the registry changes uh if we want to confirm we can fall into the registry port and see that the configuration changes have gone in, uh let's forward to a shell. Indeed, one of the registry ports.

A

If we look at the configuration file it that Etc Docker registry config.tml, we can see that etfd database related configurations.

A

With the password the postgresql host is automatically detected from our running poster scale, port and the port, the user. We specified the password business thread. Ssl mod is disabled, so everything looks right. Let's, let's come out of the port, now to confirm that if it is working as we intended- let's just by logging into our instance.

A

For that we need to find the find the password for.

A

Initial root password, which we automatically created.

A

Okay, so we are in the instance: let's try creating a sample project.

A

Okay, we have a new project created.

A

And the container registry it's empty as we just created a new project. Let's try. Let's try pushing an image to this registry or we need to login root and the password for the root user.

A

I am going to tag the adminated image to this data speed.

A

So we can see the push going through.

A

Okay, the image has been pushed. If we refresh the container registry page in UI, we should be able to see that.

A

Okay, now to ensure that the metadata database actionable, let's try, let's check what the contents of the database that we created are.

A

Let's fall into the postgresql once more.

A

And let's, let's get a PS Guild shell, but this time let's use the industry database.

A

And check what all tables are there? Okay, it has voltage tables.

A

Let's see if the tag they put ended up in this table.

A

Id1, that's a latest tag which matches with our project.

A

So the project has the idea one. So it is that product that the devil is so yeah. uh The metadata database is working as expected and the configuration change we created work file. There were no errors with migrations or riches reports to come up so I'll I'll. Stop sharing my screen now.

A

A

Image, for some reason it's not working but anyway so yeah. uh The demo is essentially this. How you specify newer configuration to your values to tml file and have you apply those configurations to a running gitlab instance so that you can test newer your functionality and see the action?

A

So with that the demonstration is complete. uh I'll see you again in the next demo. Bye.