GitLab Environments Team, 29 Jun 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Support `environment` keyword with `trigger` keyword (Parent-child/Multi-project pipelines)

Description

In this video, Shinya (Staff Backend Engineer, Environments group) demonstrates the PoC of the epic https://gitlab.com/groups/gitlab-org/-/epics/8483.

PoC MR: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/124562

0:00 Intro
0:53 Why this feature is important
5:23 Demo

A

Hi MCA staff, bargain engineer from airbrush group today, I would like to demonstrate this feature. Support environment qr2 is trigger keyword, in other words, support environmental keyword for parents, child or mate project Pipelines.

A

um Here's a brief summary of Love CM yeah.

A

Deploy deploy, job, there's a trigger queue, it is defined to trigger Downstream pipeline uh and then Define also define environment cured, to specify that which environment is a downswing pipeline going to deploy. Currently this is impossible to do that, uh but this epic is to enable the future enable the future, but before diving into demonstration. I would like to briefly explain why this feature is important um so that um you can see a bunch of feedback in this epic.

A

That, uh especially this feature, is important for large organizations um that uh prefer separate responsibilities between developers and operators.

A

So, for example, um here's a project application project that where developers are building coding and application, so, for example, developer, creates a major request and they're running a test on the building image and they're, pushing the docker image to continue registry um and did they start environment it could be review environment or staging production, um but uh open additional one to um uh trigger Downstream pipeline.

A

In this case, instead of letting developers to fully build the instruction infrastructure, there may be some compliance that organization prefer, for example, they prefer a specific Cloud, AWS or gcp, or maybe there are other um license requirement or observable abilities, Tech stack or something like that. So basically they want to regulate this infrastruct infrastructure stack in a single project. So here's a the other project.

A

This is a deploy deployment project, uh that's basically uh responsible to deploy the application project, so uh this diploma project pulling the container image from um application, application repository and then uh doing a bunch of uh uh a bunch of other processes uh for some help me installs.

A

um If they depend on Helm um or it could be like CTL or um gcp command, DCP, CLI aww, so whatever it is, um there are some toolings uh exist in this project to uh to provisioning the new environment and then at the last. uh The result is feedback to uh this deployment development project, so that developers uh can actually access to the environment without knowing underlining um infrastructure tool links, and then this is uh actually um what do we are. What gilab is interested in? Currently um this is a architecture diagram for Runway.

A

um So this is run. This Runway is a project to um they basically construct the environment for developers, so this service project is, um let's considered as a application project the developers working on it, building a Docker image, something like that and it just triggers Downstream pipeline in deployment projects so that it's uh executing a bunch of tools to uh actually making it happen, including observabilities.

A

um This actually Architects. This architecture could actually vary per Organization, for example. um These are another article about the atlasian how the approach um assertion has an internal platform as a service that vehicle micros Etc to service certain environments that enable attraction Engineers to deploy and operate services in AWS as quickly easily and safely as possible. So this is a keyword that internal platform as a service.

A

um Yes, this is a uh actually in the goal of this epic, um but this epic is specifically targeting this connection, service project and uh application project and deployment project. So without the further Ado I'm going to demonstrate.

A

um Let's take a look at that stream project.

A

Let's see ammo.

A

There are two stages: deploy and Cleanup in this demo I'm going to demonstrate how to deploy um a rugby environment, but the same thing same practice can be used for uh production or staging, and here's a diploma, this extensor uh some configurations for Downstream pipelines.

A

First of all, this is the most important word that triggers Downstream pipeline in the downstream project is let's consider this is a deployment project and then uh there's a bunch of options, Branch strategy and the variables it's important that um giving a context of Upstream project to Downstream project so that the deployment project knows which environment they should deploy to or which project uh requested the deployment.

A

um And then this typical review job has an environment keyword. This is uh the currently what's currently impossible, but uh what are we going to support? So name is a environment name review. Slash uh could be future Branch name so like a review, slash or some future or a bug fix, or something like that.

A

So let's try to create a modular cast.

A

A

I'm going to add a new feature and then commit to New, Branch um or some future one and then create a modular cast.

A

Okay, awesome future uh some future one.

A

And it is already created um pipeline. um So let's take a look at the pipeline. This deployed Refuge off trigger Downstream Pipeline and which contains problem job.

A

um It's currently uh the inside there's nothing but the um uh yeah I'm going to walk you through shortly so um yeah. It's successfully executed the downstream pipeline in the context of review this specific review, app review, app environment.

A

Now, let's take a look at the downstream project.

A

Downstream project: here's a giveaway, CMO computation. This programming job is executed when um Upstream environment action- equal is a start. um This variable was a passer from the Upstream variable as a context and also it's executed. This condition uh is allowed to execute this job only when um uh in multi-project Pipelines and then there are a bunch of scripts.

A

um This could be um maybe executing a budget deployment tools or executing another Downstream project to uh for for more complicated, um more complicated walkthrough. um Then let's go back to the job.

A

uh This job block um that the Upstream context is correctly passed to the downswing and also it's impossible to execute um API for the Upstream environment, so that the downstream project can update um the environment properties. For example, dynamically generated URL can be said after a job is done, authentication can be used here, so this is great um and then let's say this modular case um is marched.

A

um So when this module gets submerged, it's going to trigger uh another Downstream pipeline to tear down the environment and the screen app job is executed.

A

um Yeah at a similarly that context is passed to the downstream pipeline so that the um yeah the project can actually release that resource at the last. uh Let me walk you through this environment. Page um environment page works. Similarly, um it's a current uh with a current feature set.

A

um For example, we see in production environment that there are a bunch of deployments and succeeded. This all um executed the downstream pipeline, um so users can jump to the uh specific um diplomatic Pipeline and also the rollerbock is possible which execute the downstairs Pipeline with the previous Shaw.

A

um So um yeah, uh that's pretty much it for now. um Please, let me know if you have any questions, um especially, we are looking for feedback.

A

If this is uh future your um expecting or uh uh like something different um yeah, please leave a comment in the Epic uh and yeah. That's all thanks. Bye.