Flatcar Container Linux Tech talks, 12 Mar 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kinvolk Tech Talk - Pipeline Driven Development with a Closer Look at Tekton with Cari Liebenberg

Description

Cari spent the last few years building projects using Jenkins, Azure DevOps, AppCenter and Gitlab. After learning about Tekton and building a few example pipelines with Tekton, she started rethinking pipelines and the existing patterns we use to manage automation. Reviewing well known pipeline structures like CI, CD and Gitops, she discussed ideas for creating adaptable pipelines that behave in a modular, reusable way allowing projects to prepare for the rapid transformation of the cloud environments we build.

A

Cool hi everyone. So I'm doing my talk pipeline driven development and tecton.

A

Why am I here? I had some downtime between projects and.

A

I played around with tecton and I spoke to brain in a bit and I was telling him.

A

All these ideas- and he was like hey- come to a meet-up talk on it, so this is kind.

A

Of some of my thoughts and my findings on tecton stuff that I've experimented.

A

With and would like to still experiment on automated a ton of manual tasks that.

A

I was struggling with and I really liked the granularity that pipe that ticked on.

A

Gave for building pipelines- and I like that it was kubernetes natos, I'm kerry.

A

Lehman burg, I am a cloud native engineer at container solutions. I've chucked in.

A

My github url at the bottom of the slides it's there still carries it a if.

A

You want to have a look at some stuff so by day I do coding things I like to.

A

Learn I like to innovate. I hope I come from a development background. I'm.

A

Catching up on the ops there I like to do cloud native things and all things.

A

Cloud and playing around with kubernetes and doing much of their things and then.

A

By nights I play both adobe and sometimes I try skateboarding. It doesn't.

A

End so well, but it's still a dream of mine to figure out how to skateboard. So.

A

Kind of the sort of all of the swim was the devops journey and kind of.

A

Automating the pain points on projects helping the teams. I worked with to.

A

Automate those points, the desire to keep project stable, improve releases, make.

A

Releases faster, but still keep the stability and bridging that gap between.

A

Devon, ops that are brought up and sorry and the collaboration between.

A

Definitely swimming and then helping out to some projects in general that needed.

A

Some tlc cool the major thing they came across trying.

A

To get rid of those silos with devon ops collaborating with each other on a lot.

A

Of projects I worked on what we, what ended up kind of happening was in the.

A

Attempt to get rid of those silos, we kind of created devs that tried to do a.

A

Lot, they kind of lost the focus in terms of weren't able to focus on the code.

A

They also tried to take on the ops role, so what I like about tecton is, I think,.

A

It can kind of bring a bit that abstraction back dev sticking with dev.

A

Focused stuff and sticking to their strengths, and we could perhaps use ops.

A

To abstract a bit of the complications behind pipeline management, the manual.

A

Effort has also been improved through all sorts of pipeline systems, so not.

A

Just tecton, but it's been great, to see how we're improving on manual effort on.

A

Projects and then automate all the things. That's also worried here. So I'd.

A

Like to just go over some of the existing pipeline structures,.

A

Cic dcd crops and automation, city being continuous delivery and then city being.

A

Continuous deployment so both layers, it has been a very important day or two of.

A

That version control in our code has been it's given us a lot of control in.

A

Terms of the stability that we're craving branching strategies to feature.

A

Branches, environment, branches, different things, depending on your project and.

A

Your projects needs being able to work on code and isolation and then merge.

A

To the central sauce, that's been very beneficial and then other strategies.

A

Like good flow and get to ops, which I'll get to continuous integration, taking a.

A

Lot of the manual steps that we do to keep the code quality, automated tests.

A

Code quality scans, any kind of builds to make sure knowledge or breaking changes.

A

Get put into the code all of these layers, not what I made it and quick.

A

Quick feedback loops to make sure that, if anything breaks but yeah, preventing.

A

Any breaking changes getting out of production, basically and then continuous.

A

Delivery, one of the things that come up with continuous is every small, frequent.

A

Changes to your releases, making sure you're not having a big backlog of.

A

Changes that you then throw into production that could then cause a lot.

A

Of stress, it's the confidence in your releases, making sure that your release.

A

Encourage you are comfortable, bull, run and production and releasing to your.

A

Different environments, if that's part of your project strategy and feedback loops.

A

Continuous deployment getting our code into production automatically, which is.

A

Great there's it's very beneficial, creating those mutable releases. We know.

A

That that is the release at that point in time being able to deploy without.

A

Human effort, which is great, then get ups, which is kind of new, depending on how.

A

Long you've been working with it and originally coined by we've works. It's.

A

Leveraging what github or wall gets or not get up.get actually already gives.

A

You it gives you that commits pull requests, merges release tags being able.

A

To leverage that, in a more powerful way and one of the one of my colleagues.

A

Described it as a commit triggering everything or the commits being that.

A

Star point that triggers your pipeline, another colleague, believes a little bit.

A

More than that does a bit more focus, it's got to do also with how you intend.

A

That commit to end up, so you make a commit, for example, in a repo dedicated.

A

To managing your operations there, your infrastructure and that commits in an.

A

Ensign and result something else, but basically no one goes in and tweaks your.

A

Environments manually anymore, now, scp and like tweaking around creating.

A

Snowflake servers it's now this central source of truth of what you're expecting.

A

To happen somewhere else, so almost like states, management from gets layer and.

A

Then that also gives rise to some other strategies of using the commit hash to.

A

Tag, your images that you always have the point of what image is running. You can.

A

Go back to the courage to map up that. What you're expecting to see there and.

A

There's been some other cool like having that ops repository instead of checking.

A

Your operations they're inside of your project, sometimes it's a mash-up of a.

A

Project with your opposite or see on your cd with the application layer and.

A

And get-ups could kind of create a different strategy for that and then.

A

Just automation in general, the little your tools that we use to manage pain.

A

Points on our infrastructure or just in the journey investigating things and.

A

They settle your tools that you might have in your computer and you can put that into a pipeline and automate that set it up with some sort of a trigger.

A

And event trigger that reacts just things that happened in your system.

A

Where, normally you would go in and you would run those scripts manually, we can.

A

Now, actually kind of state up triggers for that or those scripts configuration.

A

And code and infrastructure as code, giving us that ability to store state.

A

And get infrastructure as code affecting your infrastructure layer that allows us.

A

To the infrastructure can be controlled from a code layer and then configuration.

A

As code form or for the application layer, we can store our applications as a.

A

State and get which is great and there's with kubernetes before example, was.

A

Stolen yamo scripts and then with terraform patrick.

A

Hope configuration language was a little bit like jason, but not quite like jason.

A

Okay, so it's some existing pipeline solutions, there's all systems, I've.

A

Already booked pipelines with azure devops jenkins get lab, get lab before.

A

The cloud I haven't both get lab locally on our local custom yet well, it can't.

A

Custard that I've controlled, but my main effort has been on get that cloud and.

A

Then recently I've looked at github actions and then I just really wanna.

A

Mention some different models: the push versus the pull model, so the lineage.

A

Goes back one step I just want to mention so both there's your source and.

A

Get and then there's a domain of trust and that's what do you control? That's.

A

What you know what's in that domain of trust, you know you've set up those systems. We go. Team has set up those systems and you you can call it the main.

A

Of trust, because you trust everything is there based on what you expect to be.

A

There, in the push model, kind of changes, get pushed to your domain of chess, so.

A

You've got you're trained to sit in a get somewhere and get repository. Your.

A

Changes get decided on perhaps in a vm, that's penned up, not by you, you don't.

A

Know how this vm has been set up. You don't know what this is a it's a bit of.

A

Like a black box, you know: what's gonna come out of it and you hope that it's all secure and you hope that is what is agreed upon to be there and best.

A

Practices and everything, and then it ends up effect or domain of trust. The.

A

Pull model was making the changes sits within your domain of trust, so you've.

A

Got that it repository is triggering something inside of your domain of trust.

A

And that then, ultimately creates the changes and so some examples of the push.

A

Model github actions which sits externally on github good lab in the.

A

Cloud which you haven't set up some examples of the pull model tecton.

A

Which I like a good lab, if you happen to have set it up within your own domain of.

A

Trust jenkins and this agro, which is more continuous deployment, focused but.

A

It's still within their main of trust, okay,.

A

Getting to take on take down now what is tecton. So it's a cloud native solution.

A

It comes from the cane native project it's being taken out and put in its own.

A

Project now it's a cuban eddie style resource for declaring continuous.

A

Integration and continuous deployment style pipelines, but you can do so much.

A

More with it, it has a dependence on kubernetes.

A

So you can't just set it up somewhere without kubernetes. It is a crd customer.

A

Resource definition, it extends kubernetes the concept behind it behind.

A

Tecton is that your patterns become first-class citizens, and it follows the.

A

Pull model that you can set it up within your domain of trust, so just quickly.

A

Going over the crd phrase, the customers see rds, allow you to create custom.

A

Resources and custom resources extend the kubernetes api. This means of using.

A

Series it allows you to use existing communities ways of doing things. You can.

A

Use the kubernetes verbs if you control, get pods or whatever it you new same.

A

Thing with the tecton resources, if you keep control, to interact with your.

A

Tecton resources; okay, ticked on pipelines; I'm just gonna, take.

A

You through the tecton pipeline's project, there's also tecton triggers and.

A

There's like a couple of other projects, they've created, but I'm gonna focus on.

A

The tech on my plants, so this six basic parts detect on basic types, the step and.

A

The step is like the commands you want to run so, for example, you'll tell the.

A

Step word image to use you give it at the stepney.

A

You can use the command approach there. I've got a cat example and you can use.

A

The scripts approach- and so you can put it in a queue of shell scripts, three.

A

Bash scripts and whatever, and that kind of gives you you can choose to go.

A

Declarative or imperative like it gives you that same flexible. If you would have.

A

With some other pipeline systems, the task, the task is actually where the.

A

Steps look, so the steps are a list of steps that you want to do happen and.

A

Then the task is like the parent of it. The task will run as a part in cuban 80s.

A

And steps will run as containers in the task pod on the ribbon at ease. Okay, this.

A

Tasks will just be resources that live in communities and you can update them.

A

You can, you can use them repeatedly, but the task runs trigger the task and they.

A

Have to be uniquely named, they run once and if you want to run a task task again,.

A

You have to create a new task run with a unique name against, so what I found.

A

Worked really well is using name time stamp as naming convention, so that you.

A

Can always have that unique task run, but do you generally weren't create tasks.

A

For yourself I'll explain in a second: why? But just to show that you have this.

A

Pick and then toss graph and your sauce run will reference the task you want to.

A

Run then, you've got your pattern, your pattern that lists your tasks, they're a.

A

Sequential order of tasks you want to run and yes and what's kind of great as.

A

You pass in your resources to your pipeline so, for example, there you it.

A

References, a github repo, and so that becomes a resource that you pass your.

A

Pipeline to use, which means you can use a pipeline for multiple repos.

A

This is what you would actually interact with a pipeline run so same with how.

A

Tusks needs to toss brent. You triggered the task. You need a pipeline rent to.

A

Trigger the pipeline, but then that's going to create a ton of tasks runs.

A

Automatically, which will then trigger the tasks you build pipeline, runs same.

A

It's got to be uniquely named pipeline run. You can use, create oscar rinse.

A

Yourself, if you wanted, but this is kind of the layer you would more interact.

A

With generally, you could actually build a tool that would trigger this, so a tool.

A

That automatically build your llamó for your pipeline run to then trigger your.

A

System- and this is kind of the level we actually pass in your resources and your.

A

Inputs, so everything else will be like kind of like a concept and.

A

Statically-Typed and there's more just schematic this- is we actually pass in.

A

Your values so that it can be used amongst your pipeline and your tasks and.

A

Then this is what I'm most excited about. Is the pipeline resources the I.

A

Mentioned before that, you can pass in multiple different github repo resources.

A

To different pipeline instances, so what this will ultimately allow is.

A

That you can have one pipeline and you can have ten projects with the same.

A

Steps that you want to run, you just point it to a different repo, and you can.

A

Point it to a different image as well, so even your images are like resources that.

A

You point to with details: there can be a google cloud image source, it can be a.

A

Docker hub images also can be. What do you choose be cool, so you can set up.

A

Continuous integration with tecton you can do continuous delivery. You can do.

A

Continuous deployments you can do get ups if you use tecton triggers. I haven't.

A

Gone over that here's also a cool talk and a cool thing to look into once.

A

You've played around with taking on the pipelines.

A

One thing I did make a note to mention here was I kind of see the benefit of.

A

Tax on not necessarily now there's a lot of groundwork that has happened if you.

A

Choose to work with tecton, it's it's nothing! You have to make it into.

A

Something like it's you're not going to get something out the box. It works nicely, but whatever 4c is kind of like the same way that there's home-and-home.

A

Trots and there's predefined things that a community has put together a kind of.

A

Imagined ticked on eventually getting these base images, cific, based pipelines,.

A

So cool at that, like pipelines, that someone else has gone and worked out as.

A

A best practice and we can kind of all contribute to as a community. So like a.

A

Library of pipelines and approaches and templates everything so tecton is still.

A

Very new and still a lot of ground work to do but like it could really grow into.

A

Some cool projects, so I did try and ask myself like: why would I rethink.

A

Pipelines like we've got all these other existing tools. There are already industry best practices. We've got, we've got jenkins, which most of us have.

A

Probably, who worked with and there's like a good lab is really great good, lab.

A

Is a good tool and github is good good to have actions. It's not a good tool as.

A

Well, not they've released it, but tecton is trinity's nature, which means we're.

A

Using kubernetes knowledge to manage it, that could be a pro or con depending on.

A

Where you come from, but I think it's a privy from really got a team that it's busy learning kubernetes has to maintain kubernetes.

A

And manage kubernetes now also has to manage pipelines for teams. They can use.

A

Their existing knowledge to kind of manage with tecton. It's not like you.

B

Have to learn groovy script like you would have with jenkins.

A

Which I mean some people might really like learning ruby scripts. It is fun.

A

It's the pool model, as I mentioned before, you can set it up quite comfortably within the pool model. Your domain of trust.

A

I like the flexible architecture, so images or resources and your cast your.

A

Pipeline resources can be images, they can be github repos, but they can also be.

A

Clusters, your clusters can also be posted as like just an object and then.

A

So you can the same pythons! You sit up against different clusters that you've managing when that has that kind of, like a very.

A

Very scalable thought process behind. If you think about it, and you can also.

A

Separate what you need by namespaces, which I'll also bring up there. I.

A

Mentioned before the idea that you can have a pipeline and user across many.

A

Projects and that's what I'm most excited for is that if we use this model.

A

Of pipeline per project- and I say like pipeline per application, if you've got.

A

Ten projects, each with ten pipelines and you need to update them. It's a different.

A

Strategy, like I'm, not saying it's a bad strategy, it's just different ways of.

A

Looking at the same problem in different ways, we could tackle a problem or.

A

Different ways we're going to tackle this solution, one from the positive side.

A

So it's just different solutions that we can bring to projects and the.

A

Reusability, so we can kind of reuse pipelines amongst teams and then I.

A

Brought it before, like the manual effort and we've abstracted, we've kind of.

A

Gotten rid of a lot of the manual effort in our projects, but now we're adding.

A

Mental effort, so it's not to say that people don't want to the metal effort.

A

It's just: we must also be careful not to keep adding mental effort to projects.

A

Like it keeps being a new way of doing things and our bringing up take on us a.

A

New way again, but kind of we need to stabilize on the mental effort that.

A

We're putting into projects- and I think, if we have this idea of bringing focus.

A

Very developers perhaps set of projects where they're just passing in the.

A

Pipeline resources but they're not managing the pipeline per se. The pipeline is already pre managed by another team. I think we can listen. The.

A

Mental effort- that's what's happened, it seems, and maybe we can get those unit tests written, because that would be great and I think we can kind of get.

A

Innovation in the ops layer, where ops get to be creative and they get to think.

A

Of pipelines that are reusable in a different way again, it's it's just.

A

A different way of looking at it and we bring that fergus back for this.

A

Just some more cool things that I taked on that I liked I didn't really point.

A

Out inputs properly before, and I like that, you can, from the pipeline run side.

A

You can pass down an array of values, for example, if you have a command with a.

A

Couple flags you can pass in those flags as an input on your pipeline run. You can.

A

Also pass in string values to be used like, for example, the url that can be.

A

Used in your pipeline in passing up from that pop and run there, and it will go.

A

Down into the task layer, some resource types, I've mentioned the get resource.

A

There's quite a few and it's still being added to and improved upon. I, the image.

A

Resource and the cluster resource this you can also use storage, storage.

A

Resources and the cloud event resource is what tecton triggers uses to allow.

A

For the guitars there and then you can also output things but artifacts between.

A

Steps, if once it needs to output, sometimes you need to use in a different.

A

Step that they fall so keep that in mind with being able to share artifacts.

A

Between steps, its multi-tenant there, the reason I say that is, you can control who.

A

Sees worked with our back and name spacing I'll show you in a second.

A

Through the detective on the dashboard kind of how you can kind of, if your team.

A

Only has access to certain namespaces, that's all they'll, be able to see if.

A

Your pipeline is running on those namespaces, your pipelines and your.

A

Tecton resources that you've assigned through those namespaces.

A

And I do think ticked on comes within the devops vision of trying to create.

A

Collaboration between the teams again, I I think it brings back the thought of.

A

Collaboration and not just trying to overload yourself with a lot of tasks.

A

Trying to take control of your tie, your project from dev to the offside I.

A

Think um I don't think it brings back silos. I think it still brings back.

A

Collaboration and back to that spirit of I do my thing while you do your thing.

A

Well, let's work together and we'll create a great project together, so I'm.

A

Just planting a seed- I I don't know if it's the final solution, yet I mean there.

A

Might be a new tool soon, but just so I mean the day you plant, the seed is not.

A

The day you eat the fruit, so I'm just putting it out there. I think it's a cool.

A

Solution, that's we can do some cool things with it and don't fix something.

A

That's not broken if your team's work well with ci and the cd built into the a.

A

Player and they understand it and that works for your team. It doesn't have to.

A

Be changed ticked on it's kind of. I saw it as a solution for square. I saw that.

A

Wasn't working so I'm already thinking of ways that I can use tecton to solve.

A

Things that haven't been solved yet cool and then references.

A

I've set up a repo ticked on examples and I kind of keep adding stuff to.

A

Github, that's also why I keep mentioning my github account, because when I figure.

A

Things out, I kind of to share it on there, and I just kind of like now and.

A

Then I'll do a blog post, but I mostly actually just throw my examples on.

A

Github cool, thank you. I'm gonna try demo! It see if I could record myself. So if.

A

This doesn't work so great. We're gonna watch a very long recording. Okay, let's.

A

See I'm gonna try murder my screen. I think that's going to be the best option.

A

Here, if I'm a mouse, I found my mouse and I'm just quickly.

A

Finding my screen settings once again.

A

Bear with me, okay,.

A

So gonna try zoom in a bit here so right now, I'm on the default namespace, alright.

A

Okay, hold it I'm just quickly. Let's make this big.

A

Wow- and I really made the one thing- big sorry, technical malfunction set a little.

A

Bit there: okay! Stop with that. So I've got my tecton resources set up on the.

A

Tecton pipelines, namespaces is just custom at the box grab my pods, so I've.

A

Set up the take on dashboard, also just following the the tutorial. If you do go.

A

To my tecton examples, I've done ascii cinemas recordings of these tips and I'm.

A

Just really going to port for it, the stream interior, I'm just gonna, put in a.

A

New tab, quick, this okay and I'm brain can't remember, come on.

A

It's just bear with me: okay, it's a port word is namespace. Okay,.

A

It's stop breaking. Obviously I had it all set up.

A

Okay, so let me just quickly grab my dashboard back. Was it the second one.

A

There we go: okay, no or 1997; okay,.

A

Sorry guys it's gonna get better soon. Brainy will get better. Okay.

A

Okay, we have things, this is sticked on.

A

By no I'm kidding okay, so this is just a quick little dashboard that that is it a.

A

Sweet gooey, it's not as great as some of the other gooeys. I think they're still.

A

Working on it, I like it for the visibility, but it can still be improved.

A

In terms of its ui, I've got all my plants on the demo namespace, so this is.

A

One of the things I mentioned about like namespace, if I didn't- have access to.

A

These namespaces, I wouldn't have been able to view them. I've created a little.

A

Ci pipeline that I've been using and I can go into it- I can see I've got like a.

A

Linting step test step some coverage and then I do a bolt. I'm gonna quickly run.

A

The pipeline now and take you through the resources it takes about ten minutes.

A

So I might not be able to show you the actual bolt, but I can show you pause.

A

Bolts and just trust me that it's going to get there okay, so first we open up.

A

The pipeline itself, so I had a private repo that I went to.

A

The test checked on poppins against because I got.

A

Public repos working fine, but what I really needed was a realistic private.

A

Repo scenario: that's still: digital, okay, okay, so here's my actual pipeline, I've.

A

Set up its see on our j's, it's a no j's project that I set up somewhere. There.

A

Tried to run a pipeline against to see if it can add value to an existing.

A

Project that I've got I pass in the path to the docker file for the koneko bolt.

A

Path to context and an image tag, I'm hard cutting the image tag right now. I'm.

A

Not doing like a fancy, he commits, I think it's just hard coded for now, a.

A

Parson, the github reaper and the gcr image details the the resources I've set.

A

Up it's a private reaper on our company repository, I used github authentication.

A

Details to get in there and then I've got a google cloud. A google container.

A

Registry details, then I used google's service account, jason, the credentials to.

A

Set that up and then a split out steps for myself, a bold step that will bolt.

A

The image using koneko so there isn't root access happening and.

A

Run some coverage tips code coverage code linting and run the tests.

A

So I'm really going to trigger this using the pipeline run.

A

I've got customized in there just to kind of solve the consistency between.

A

The namespaces for me and yeah, so I I kind of do this manually right now, but.

A

There are better ways to do this. This is just the prc of a plc, and so you would.

A

Apply changes because this is a pipeline run it's unit. It's got a unique name.

A

That's why I had a manually change it quick, but you can put tomato. I'm.

A

Applying it with the cake flag so that you can use the customizer just like you.

A

Would with other communities, resources are gonna use the namespace? Sorry hey.

A

Hold on hold on, I know it's, okay, this actually the runs folder. Okay, so now.

A

It's created the pipeline run. The pipeline run is triggering checking your.

A

Pipeline and then it's going, okay, cool I've got all these tasks. I need to run.

A

And I've got these steps in the toss, so it's going to spin up a part for each.

A

Task and it's going to spin up containers in each pod, so I go. Let's use.

A

This guy, okay, so this is the other thing there's some.

A

Like log rotation, you'll have to build into the system. If you use tecton.

A

Because it really does clutter up like the data, but I've kind of just been.

A

Going what was it? Was it okay? So it's got its got. Your steps go.

A

Back quickly to the code- and I go to my ci pipeline- see here- I've named it like.

A

Step one nerd jslint step: two, no j's test step; three, no j's coverage, so it's.

A

Got those same corresponding steps here, the the lindt, the tests, the coverage, the.

A

Bolt I can interact with these similar to what I would with pods. So if I want.

A

To investigate what is happening in the pod, I can, for example, just grab the pod.

A

I'm just going to spit out all the locks from all the containers and all kind of.

A

Okay, I should have saved that lisa, but yeah. Okay, there, okay, okay,.

A

So now I can do that with all of the parts. So it's about all of those four.

A

Parts fit out all the logs, but I'm going to jump to the gui because that's just.

A

Gonna be visually easier for us, so yeah we can see our the pipeline run that I.

A

Just triggered is still in the process of running and the different steps.

A

Except for the bolt we're just going to take about ten minutes to finish off, but.

A

Yet so that you can see the logs here, you can also see details of the tasks.

A

That you ran, which is handy if you're debugging, and you want to see what.

A

You've done this, some like code coverage there just sped out just for just to see if I could do this kind of stuff with tick-tock and.

A

Yeah, so what will then happen, and I'm not gonna make you such a for 10 minutes.

A

Watching the things been, but it's going to tag.

A

Kaneko is going to build a container from, I mean an image for me. It's going.

A

To tag it with the tag I provided and the params.

A

Which, like I said it, can be a commit hash. It doesn't have to be hard-coded.

A

The way I've done it and then it's going to end up here. It's gonna push it here.

A

Which, then I can use right wherever what I wanted to do. A continuous deployment.

A

Demo as well, but time happened, life happened and I hope the that part my.

A

Kind of shows where it can lead and where it can go yeah. So thank you. Thanks.

A

A

I will be around if you have comments, questions.

A

Any questions I installed it a while ago. This is quite a simple, install.

A

Process, I'm quickly, gonna pull it up to sir and mr. critz. It's I do think it was.

A

Something as simple as like running a home command.

A

It's just quickly have a look it's, so this is the actual repo and then, if you.

A

Quickly go to the installing tecton pipelines screen and then yeah, so it.

A

Creates a cluster creates some of the are back stuff. The casserole bindings.

A

That you need and then yeah it looks like it's just a young old script, not.

A

Even home yeah, you just run out of yellow script.

A

And then the dashboard is pretty much the same as well. If you want to add the dashboard to your system, you go find the dashboard repo. It also just heard of.

A

Mine, so second question is: when you run these tasks and black lines and things.

C

When they finished executed the pods up, we should start with the containers.

C

And sort of complete, so they'll run as jobs inside I've also want to find out.

A

Why why are they just running as pots and then they stop? Because you can do.

A

Almost similar things with jobs- and you can still have your audit ability with.

A

Jobs, so I'm not sure why they chose past like jobs. I still want to investigate.

A

That actually the tasks, the tasks seem to run at the same time, but the steps.

A

Seem to be sequential so like step, and then it goes into the.

A

Next thing- and the next thing I could be corrected, but that's what I've.

A

Experienced so far so yeah, I think that's how it kind of has been found. So.

A

Your tasks will kind of run parallel to each other. Then this you've explicitly.

A

Stated that you want them to wait, there are some other cool things you can do in the mo. You can put conditionals in there. Like only do this. If something else has.

A

Happened and stuff, and I think you might be able to do stuff like controlling.

A

More explicitly the the order of tasks, no, it's still a tech time. So it's the.

A

The same team, that's running, tecton pipelines is running tecton dashboard.

A

And it's running tecton triggers I haven't seen any other. I mean that's.

A

Community involvement, so it's open source so whoever, but I think that's.

A

Still, driven by the same people.

A

Well us debugging stuff, I mean I did some stuff. That was like probably not.

A

Best practice like commenting out tasks, so I could isolate the tasks and just.

A

Find that one task it's not working, but I was thinking with the conditional.

A

Stuff we can start adding if in debug mode run this with that, like, I think, if.

A

We start wanted to, I don't know if that's gonna, add more complexity and.

A

More clutter to it, but we could run that, for example, in it up, like not noted, like.

A

A devops environment, not the devops phrase. I just mean like it's our.

A

Development environment for operations, procedures and we can test it out before.

A

We move into actual ops layer where we could have two pipelines. I think it also.

A

Was something I was excited to explore further, I haven't debug the too thoroughly yet in terms of that, I did do a couple like.

A

Echos and my tests like that, but those aren't I mean they just they probably.

A

Made my life more difficult than easy, but I think we can strategize with some.

A

Nicely thought out: debug steps, essentially docker containers inside.

C

Human age right so, if you've got a stick, all the lint does something you could.

C

Literally around that same time, feeding into it, there is no unity, whether it's.

C

Incubating so much easier to debug, because if the stock makes it all nice.

D

Elation, sorry, yes, so you know you do always have this.

A

Looks that aren't get destroyed, they stay there.

A

Unfortunately, so you're gonna have to do some cleanup eventually, because it's you.

A

You can see it becomes very verbose, the more steps you have in the masses, I've.

A

Been a little we're just gonna have all these like logs collecting everywhere so.

A

We'll need to be some sort of strategy flat.

A

The only thing I can think of to do that.

A

Which is a very nasty way of doing it is to put a sleep in there and just keep it.

A

Awake but I haven't seen anything that will I don't think ticked on designed it.

A

To be able to do that, like I don't days, I think the intention of me lifted that.

A

The one support is, then, it's done, but I think they give you in that freedom that.

A

If I want to force us stay away, if you can put one or two steps and commands.

A

And that are like a little bit like hecky, but they kind of help you with.

A

What you need it for things that you running hundreds of pipelines a day.

C

Would you, then, just to make some cleanup which should remove some of those jobs.

A

So because, if you, if you keep running jobs and keep your any pipelines, it'll.

C

Keep building up and building up and every time you do get parts it will be.

C

More and more and more, would you just do cleanup on that to go keep the last 10.

C

Or 20, or keep the last yeah three days, so you could set up stuff like that log.

A

Rotation was something I wanted to look into next. That kind of strategies for.

A

That in general, just feel like kubernetes maintenance as well. I.

A

Wouldn't I would move it somewhere else like to an archive. I wouldn't anything.

A

Cuz, that's that's the order to bility, like that's. That makes you high any.

A

Changes auditable, if you get rid of those, then you've got no audit of what.

A

You've been doing on your system, so that's the only thing is yes, you could.

A

Set it up, just don't delete stuff, remove it, something like our cabinets.

A

Partners can install each cluster and you can say like a. I need, keep this job.

B

What is your last 24 hours, or so only this cannot always lock rotating is.

B

Called clean up clean up operator, look into that. Thank you, okay. Thank you.

C

Everybody, I think, I'm out of time.