keptn Community, 2 Apr 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keptn Community Meeting - April 2nd, 2020

Description

Agenda
Latest Bridge EAP: keptn/bridge2:20200402.1046
https://keptn.sh/docs/0.6.0/reference/keptnsbridge/#early-access-version-of-keptn-s-bridge

Keptn Survey
https://forms.office.com/Pages/ResponsePage.aspx?id=o-PrcDBbXUOdZ3cW10yhkLnGNhAC6D1Dpu2wuyXaUqJUREJGVjlCWkdTNjNZVDFTVDc2Q0VXVEE1Ni4u

Introduction Keptn Sandbox
https://github.com/keptn-sandbox

Main Act: Keptn with GitLab
Integrate SLI/SLO-based Quality Gates
Automate SLA Checks through Dynatrace Synthetic

Useful links
https://github.com/keptn-sandbox/dynatrace-synthetic-service
https://gitlab.com/checkelmann/dynatrace-pipeline

Outlook
April 9th: Second Keptn DEV meeting
April 16h: Keptn Integrated in Atlassian

Meetups / Keptn presentations
April 8th - Neotys PAC Beer (virtual)
April 21st - Comtrade Quality Quest Meetup (virtual)
May 4th - DevOps Stockholm Workshop (in the works)
June 25th - Dynatrace User Group Benelux
July 1st - Cloud Native London

Meeting notes: https://docs.google.com/document/d/1y7a6uaN8fwFJ7IRnvtxSfgz-OGFq6u7bKN6F7NDxKPg/edit#heading=h.uwwfn8xf7bjf

A

Welcome everyone to the next Cape and community meeting it's April 2nd week, 3 of the lockdown, but we're still in good spirits, right Christian. Exactly so Christian we will you'll be the main act of today's community meeting. I just have a couple of other points on the agenda, but then the floor is yours: I want to make sure. First of all, this document is accessible by everyone right. If you go to the captain community meeting page, you will find all the previous the agendas and the recordings and all that stuff.

A

So the first one, a quick fYI. We just released the latest bridge really access program. So if you hopefully know that we have on the captain's side by the way, Christian just double check, you see my screen right, yeah, it confusing perfect yeah. So the on the captain website, if you click on releases documentation and then on the references, you find an entry for the captain's bridge and he was always the latest and greatest.

A

The EAP version that we put out there and, as you can see from the timestamp, should be pretty obvious. This is really completely fresh off the press. It's from April, 2nd and I just happen to have it installed earlier, and it's really really cool stuff that folks have been putting in here. So there's a lot of hidden but hidden, but cool I like smart links, I think as they call it. Also the.

A

Let me double check here: they the heat map, so they did a lot of improvements to navigate around so check it out make sure that you are up to date on the latest version and then please give us feedback about it too. This leg channel, so that's the first thing, I think I. Actually, one of the features Krisha that you will show us later on are a deep links.

A

Y next later second thing is: we've put out a survey and we will post a survey as well in the Select Channel later on, but you may want to grab it from here as well or actually I'll copy it and I'll put it into now. I just need to find the jet window, it's typically the biggest challenge. Where is my chat window?

A

They just do it yeah folks. Can ask questions as well we'll try to moderate as good as we can and once I find the chat window chat. It I'll do it later on once yours you are presenting, but the point of the survey is: if you open it up, it is a I forum.

A

We want to know more about you and how you actually ended up with kept, and how did you hear about it? Which version are you using and so on? Just a couple of questions that we would like. So we better understand where you're coming from what your intentions are, which environments you're running on and what you want where you want captain to go. So this will help us learn a little bit more about you. So that's the that's. The the captain survey.

A

The third thing I want to talk about quickly is the captain sandbox, so you may have seen it in last week's presentation from Christian where he walked us through creating a captain service from scratch within a couple of minutes. So if you have your own projects, your own captain extensions, whether it's a captain service, whether it's a tutorial, whether it's just a bunch of scripts that help you integrate with another tool. Then please tell us about it.

A

There is, if you go to the captain, sandbox a repository, it's pinned, it's called contributing and you again didn't a phenomenal job in explaining everything how you can, how you should prepare your captain sandbox project and how we can then actually pull it over to sandbox. So once it's in st. box, because if all the projects have a place to live- and once we see this is a project, a lot of people are using it as the right quality status.

A

Then the idea is we're moving it over to captain country, which is officially kind of capital contribution. So it's we're kind of following the CNC F rope flow here. So these are these three things that I wanted to highlight and I think now the main egg Christian. Thank you so much it's about cap with gitlab and I have used your example in the past of different presentations, I've given different meetups and conferences, but now we want to hear it from the creator on how to create it, kept with gitlab and what the use cases are.

A

So I will stop. My sharing now and I see a lot.

B

A

Questions are coming in, that's great, actually, Peter Griffin, it seems you are hitting the h button because we have about a hundred questions from you may hopefully didn't fall asleep and on the h button.

B

So much green yeah, so you can steam it screen. Yes, okay, perfect. So what we have here is basically a very simple good luck pipeline with a build stage deploy. It has the verify stage on the build stage, we're just creating a little docker image which we can then deploy to our cluster on the deploy stage. The first job we're gonna run is the great captain quality gate job which will actually ask captain. Hey captain. Do you have already my project onboard it on my service onboard it?

B

If not, it will create the project in service for you and then it will also update the S line. As lowell fulson happen, then we will deploy the application tune, monitored environment from dynaTrace. Then we will send and deployment finished event and then, with the deployment finished event. My new captain service, which is called the captain synthetics service, will automatically create in synthetic checking dynaTrace for your service, and then we are running some load tests using tray meter, but this is totally optional.

B

So if you're running a meter test, you can as well use the captain che meter service for for running this test, as we in our company using some other load generators and so on. We are still relying on this stage and when the test is done, we will start in captain evaluation with a captain ablation chop.

B

This will trigger. As a captain start evaluation went in. Captain captain will then retrieves the slice from their slide provider, which is Dino tracing in our case, and we will yeah ask in intervals. Hey is stimulation done and if it's done we'll get results back directly to our pipeline and and create our yeah that deployment and stop the pipeline. If it's failing and if it's not failing it, will directly go on to the next stage.

B

So yeah, that's basically what the title is doing so, let's trigger the pipeline, so I have configured the pipeline that it should run every time. I create a new tag over to create a new tag here.

B

B

Let's see if the.

B

Yep, so the populace running in the meantime I want to show you what you have to configure in your kit lab environment or in the github repository to get these all stuff up and running. So first of all you have to connect. We could be Nettie's cluster to your github project here I'm using git Lacombe. So if you're running git love, on-prem or get love in your in your company, you can define this Cuban a disaster, not only on a project level.

B

You can define these just as directly on your groups, so this doesn't need to be done every time. So I have my kubernetes cluster dynaTrace demo with an environment scope of star, so it's a wild card, so every environment book being yeah publish to this cluster and why immense our one-to-one mappings for the kid'll Abstergo table.

B

Sorry for Captain stages, so a few mind, while I'm in test, which represents my shipyard, llamo file, yeah my shipyard yeah Mel stage, which is as well test so and last but not least, you have just to configure some CSD variables which we are using within our pipeline.

B

So we have here our planet, race, API token and our dynaTrace tenant ID. So we can directly speak with the dynaTrace API within our CI CD chops, so and last but not least, the repository itself.

B

So we have here our faults: we have our s Lionel or as oh yeah mall and our shipyard file, so shipyard file is very basic. It's only one stage name test and.

B

Yeah what I like on this solution? You can use C as a good upstream repository in Captain, where can just as line as Lowell files, but if you don't want to switch to an extra repository, you can just adjust the settings here within the two files and then, when your pipeline is running, it will automatically update the files in your captain project.

B

So let's switch to the web IDE and go directly into our pipeline configuration. So we are here to get lob cimu file, which is my main entry point of my pipeline. So I have to find here some global variables, like my project name, a service name, then I'm, including my templates, so I have my create captain quality cake. Template template my create deployment event. It's alerting. We will not covered today my to meet a load test. Then my problems, API template and my captain evaluation template in github. You can not only include local templates.

B

You can include template from central repository. So, in our case, in our company we haven't sent a central template repository and then you just reference it like this include from project, and then we have your group paths, template repo and then the file, and then you can say: okay I want to have these tech version of the template, for instance, or you normally, when don't reference here, anything it will pull the latest master.

B

So next we are defining our stages, likes, build, deploy, test and verify stage you have seen before. So we have our build stage, but that's not really interesting. So the first thing we are going to run as a cat, crate, captain quality gate create or update. So this is basically included from our great captain quality gate. Yama I will open this far. I love missed. Second.

B

Where this quality ate- and you can see here- I'm referencing the environment as well test. So we need this environment to tell captain which state you should work on. So.

B

M plate yeah, it looks like a normal crop in catalepsy eyes, so I'm have here my stage to find I have my image, which might get lucky on a run eyes using the CICE tools. Latest I can show you quickly what I mean. This was a file with the image, so I have prepared an a little Alpine docker image where I have to captain Eli installed, which I'm using in every job when I'm, when I'm working with captain so I trust me to run it with my I.

B

Don't need to to install it every time so I'm just including the stopper image which isn't in the good lab registry and then I can start using captain commands directly out of it. So first we are authenticated against our Cape captain API endpoint. So if my variables here with a captain endpoint, this is really straightforward. I think it's already in the documentation right. It.

A

B

So getting my endpoint my API token and then I'm, because captain or the captain until I actually do not provide any functionality that I can say: hey captain which services are already on board. So I have here a little workaround and clearings lighthouse config from a project and it's allowed lighthouse config exists, then I note my projects already on board it and if not, it will here, creates a project and creates a service, and you also.

A

Have referenced up there, they kept captain enhancement proposal to just yeah.

B

A

This capability into the CLI yeah.

B

A

I think it's really amazing. So what you basically do here is you. You just create a captain project on the fly in case. One.

B

Has not already.

A

Exist for your kit lab project, yeah.

B

Right exactly so, the developer itself has nothing to do so. We did not necessarily need to deal with captain seal I and and creates a service by himself. So it just needs to include the qualitystage template, a quality gate template to his project and the template will take care of the rest.

B

So if we are creating service, we are also configuring, the lighthouse config map for the project and then updating our S line as Lowe's and last but not least, we need to configure our dynaTrace secrets for the dynaTrace s live provider. So that's what the last lines here are doing yeah, so we're getting secret phone from the captain, namespace with add an address credentials, project name and if you found secrets or if no secrets is existing, he will create a new one, and if there is an already secret, you will recreate secret. So.

A

Basically, you bootstrapped, you automate it to complete creation of a project in captain onboarding, a service you're.

B

A

Or updating the s allies and s ellos configuring in this case the image weights is the data provider, but this could also be, let's say for me, feels because Prometheus and then the only thing you need to do as a developer that runs a qiblah pipeline is include that particular step and then you're good to go. Yeah.

B

He just needs to include the yamen file in now here, in this case, from the local repository, but so that it can be in an central repository. It can as well be I think in kit lab. You can run something as four stages, so you can force every proach group to to run some some chops, but there's nothing. We are doing right now and the only thing is to to add some some additional context to the to the template, like environment name, and these only text in just that.

B

The top should only run on text because I'm yeah I'm putting a lot of code in this repository for for testing purposes and I, don't want to run every time. It's a pipeline, of course, and I put coding this really into it so Oh. Next, we are deploying our application, as well as the same procedure. I have my image: I have my environment and then I'm running my helm, upgrade command I'm just doing this, why a helm or or this is only for the demo.

B

Normally, if you're deploying a workload to Cuba Nettie's, you could completely use captain. So this is, as I said, only for the demo. Here you can deploy to any other systems. If you want, you can use something CloudFormation or whatever it's just as an example how it could be done. So we are deploying our application waiting that deployments done by helm.

B

Then we are asking helm for our health, send point and our ingress host and when we have everything together, I know the exact URL from my deployment and I'm creating an unique identifier which I can then use for the deployment finished. It went in captain. So here I'm we are preparing my my JSON file, which I will and after to my to my captain deployment Howard to Mike to my captain.

B

So it's a deployment finished event. I, add my capi project pass to source, adding the ID I have here my project name and environments? Luck! The environments. Luck is environment variables, so it's test. I have my service name. My deployment strategy, mine, commit my commit tech, my registry image and then some labels and it's interesting. Post labels are here the synthetic manual, a sign up and the frequency, and, last but not least, my deployment URL euro I public, which is I, think it was introduced in captain zero. Six one right exactly.

A

B

So I will write my event down as a file, because I want to have this file as well as an MC I drop artifact then I was indicating and my captain and yeah Santi went that deployment is finished and then you should see in you captain bridge.

B

Already third deployment, which we are triggered by a pipeline, you can seize labels and, oh actually, civilization, is already done. Oh no, it's start a valuation here. It is ok, then, because captain right now with the deployment finish, the vendors first of all, it's a synthetic check creation triggered as a synthetic service, I've written as well as the deployment event is pushed to captain and to dynaTrace, but unfortunately there's some data missing right now, which I want to see in Tana trace like my pipeline CI backlink, which is right now not passed over.

B

So that's why I create another deployment. We went with another template, we're.

A

Moving these things over thanks to.

B

Your feedback into.

A

The energy service, yeah yeah.

B

Instead, it will be obsolete when the captain done free service, capacitors and deployment informations yeah. Then we are running our load test stage. I, don't think I need to show this, or do we want to see it. It's.

A

B

Ok and then we are running the most interesting stages, our captain evaluation stage. I will start it here, a little bit delayed because my Tomita test is really really small, so it will run in a minute or so and to have yeah a little bit more of a normal or authentic scenario. I'll. Let it wait for a couple of minutes and then start the evaluation. So it's a whole Asian stage. It's basically the same. You have seen before unlocking my captain and then it's getting interesting.

B

So the first thing I'm gonna do is I queries the dynaTrace API to get from my captain service, the entity ID with the entity. Id I can query sedan, a dynaTrace event API for my last deployment, I've done on the service and with the last deployment I'm getting the start time times them, which can then use to pass over my captain evaluation start time.

B

So I have my start aid then I'm getting my current timestamp, then I yeah I'm just calculating the minutes, which should be looked with relation to an event or with a happen, evaluation, sorry and then I'm, sending the start of elation event from a project and stage with a time frame which I have calculated, and these darts time so and with this command I'm getting packing a context ID from happening, which I can then use to ask the captain API, hey captain: do you have an evaluation done event?

B

I will repeat this: every five seconds till I have my relation. Thani went from captain and then I can take a look in the result. So this is basically ended in the data result here and if the result is failed, Oh. What I forgot to mention?

B

Sorry, your your deep link, I'm sorry so, with the bridge viewer, L I have my captain bridge my service name, my captain context ID and my event ID, which is also in the in the evaluation done event and so I have my direct backlink to the captain's bridge and if the status is failed, I will fail the complete pipeline and if it's finished it can go further and it will print out. So it's a pretty or if the job is failing.

B

The person who triggered the job will get an email from it lab or you can send slack notification or whatever you want. It depends on the integration you have configured in your bit lab environment, and so the developer will get an email. So hey your job failed because of L relation, and here is a link directly in the captain bridge. If you want to take a look wide spinning. So let's go back to our.

B

Ci CD pipeline.

B

So you can see everything is done and passed and he evaluation stage. I have directly.

B

First, open link here that.

A

Is pretty cool.

B

Link to the level agent on happened.

A

Perfect and then you click on the evaluation comparison and then we also see the perfect.

A

So, to kind of recap: I know you: should you walk us through the whole code that you wrote your kit lab integrations? This is also all on your github repository at me.

B

A

They want to do the same thing. They just add these building blocks into the kit lab pipeline, which means that you are sending let's say, captain a deployment event which tells captain about the deployment which can also then be used for automated test executions, or the captain can actually run the test as well, and you were doing it from your pipeline right now. I think what we have not seen yet may I'm sure you show us is the this will also trigger the creation of a synthetic test.

A

So every time you deploy a service, you automatically create a synthetic test after your your load tests are run. You also, then, do the the quality gate, evaluation, which are based on a slice and SLO study developers, have specified in their own get repo.

B

So here the it's monitor recreated, so it was actually created, but it will update all the time. So if something is changing like the like the URL of your deployment or so on, it will check paid to a heaven already in synthetic monitor for your project, service and stage, and if so, it will updates monitoring and if not, it will create a new monitor and.

A

I think the cool thing here is, if you are using pipelines to deploy services then- and we talked about this in the blog that we wrote together, then obviously kubernetes itself internally is checking the health endpoint. If you run, if you deploy in kubernetes, but if you're deploying kubernetes and your service is accessible from the outside, you have no clue. Well, no validation. If the outset link is actually working. So that's why setting up with synthetic test fully automatically to test the outside link is great in the second thing to works in application.

A

If you, you know those that are not deployed on kubernetes.

B

And if you want to take a look, the project was moved over to set sent captain sandbox on Monday. Are you so I'm? Not a goal developer I have to admit, but if you have any ideas or any contributions, just yeah contribute to the project. So there's a lot of things we can.

B

We can put into like more configuration options or like stuff like click path, scripts or so, which comes to my mind, actually all right now, when you're deploying synthetic service, our the captain service, it will it's only supporting HTTP checks, which are only runs on active gates with a synthetic enabled.

B

So it will look in your dynaTrace tenant, which active gates are available and assign the check to every active gate. So perhaps another thing I want to make it configure a blonde which active gates gates. It could be run and so on. However,.

A

It's cool its automated SLA validation. You automatically set up your SLA tests from an external like through in this case dynaTrace synthetic an you.

C

A

B

Awesome question anything else: I don't know if want to see if a get problem stage well,.

A

Yeah I think that's definitely also another interesting aspect, because you are basically here just clearing for any problems right. There.

B

A

By the monitoring tool, again in this case, dynaTrace on the environment that you, which is deploying and running tests on so you're, even extending the the pipeline validation and beyond a slice and it slows, but actually also factor in any problems that may have been detected by your full-stack monitoring tool. Alright,.

B

So um word is get problems so.

B

Yeah, it's it's just one: curl command the dine phrase, problems feed with my captain service, my captain state, and if something is getting returned by by this query, I will fail the pipeline as well and print out a link to the actual problem ID. So you can click on the problem idea on the link in your email, you get from from good lab and then come directly into your Daniel Frye's tenant and see which problem was detected.

A

Cool, do me one more favor before we switch back to my slides and then kind of open it up for question and wrap it up. Can you go back to the pipeline because.

B

A

That a large portion of the captain core team is isn't the call and I think. First of all, it's it's great that you know. First of all, you but also the captain team has designed kept and in its current state, so you can actually do all these integrations through the AP is- and it's very interesting to see that you are you're using for kept mainly the CLI, because you baked it into your into your container into your tool container. You.

B

Know I'm a lazy bastard, it's so that's: okay, okay, I am touching curl and I feel I commend I'm using this one, but yeah you're right, I I could do a lot of stuff with with captain API as well. It's.

A

Perfectly fine wettest virus Eli steer for, but I just wanted to recap, also for us to take back in terms of how can we make this even easier? That means you I know. We talked about that. You get additional comments on the CLI or the API, so we can query existing projects and existing services and all the team has already does already have this in the bucket list so that we have all the the crowd operations supported. So that's already coming. Yes,.

B

That's what we are doing here and create captain quality gate, all the stuff which I'm right now have the workaround with.

A

B

A lighthouse config exactly, but if I could query directly the captain's Eli it would be much nicer. Yeah.

A

So that's one thing: the second thing I can just recapping, because this will be great. Is this great input for our future development? You mentioned that the labels, the latest stuff that you put on the captain event, whatever event, should then also be understood by the diametric service. That is pushing these events to dynaTrace and then taking the labels is custom properties, so they also show up in the analyst. That's already. We have within an item that already on our bucket list, yeah.

B

Let's hear me quickly: load my service first configure I have my employment events.

B

This is the captain applied deployment event, and this is my manually apply deployment event, and here I have my CI back link, so I can directly click here exactly and see which pipeline was deployed or have deployed the service version and as well as directly which top of my pipeline and if this could be yea passed over in any any kind of label within the deployment finished event. This would be awesome. Yeah.

A

So I think that's definitely something we can work on away. We have no list. The other thing is your pipeline. Now you know we. We have a lot of capabilities already in captain when you tell Cap'n about the deployment, it can then automatically execute the tests and then do the validation. So- and you said this, this is part for the demo, so this pipeline would look so different and much cleaner and smaller. If you would just say, captain here's more effect applied or captain I deployed it.

A

He was my URL and then you do all the rest. Basically, the whole testing phase would be managed by captain and then the validation of the only thing you need to do is I have a Department and then it just wait for the result erode. It's also really cool. So.

B

In the end, what when we leave this stage out of the of the pipeline and running the train meter tests with the captain jmeter service, we can just adhere in the verify stage or or we can add, a job which will vary the captain, API endpoint, to wait for a finish, the wind or or at least not at least but but just what needs to be done.

B

If it's running is all the captain services, you just need to wait here on a job like the captain evaluation stage for the evaluation done event, and you know you train meter, tests or past emulation has passed and- and that's it that's perfect and.

A

I think we should also think of folks team that are on the call and listening in I like it that you are also clearing the Diana Tori's problems. We can also include this I believe in the evaluation stage, because it is just another, is Salahi metric. The number of problems that have been opened for that particular take and there's.

B

A

B

So from perhaps in the donut race, I provide our dynaTrace monitoring service yeah.

A

Exactly very cool Christian I'll just quickly take over control and just.

B

Share my screen.

A

And I want to say thanks again on the windy notes. We added the links to your repos also to your ticket. Life. Extensions and Christian, maybe be also at some point one or think of whether we want to move them over as well to the sandbox.

A

We can figure this out, it doesn't mean you can probably tell us what makes more sense where lives, just an outlook there's going to be every week, we'll have a developer, focused meeting and then a user focused meeting. So there was a user focus meeting next week we have the second def meeting and then on the April 16th. We have a similar presentation as today, but in that case it's going to be kept and integrated with the ablation tool. Stick so we'll see, bitbucket integrations and so on and so forth.

A

So was a couple of meetings coming up. Things obviously have changed I'm moving around with all the virtual events and some events being moved or canceled. So I will try to keep this up to date and I also saw there were some questions coming in, but it seems these questions all came in from either BOTS or people that wanted to they'd have too much time at home and really don't care about the problem. But what we're talking about some people were worth falling asleep on their own.

A

There keep a keyboard, I, don't know, but if there are questions from the audience that is still out, there feel free to put them into the question feature. The Q&A feature went to the chat window while we're waiting. While we give another 30 seconds or a minute on this to type in your question, you again is there anything else that you wanted to add that I've forgotten about maybe upcoming upcoming events.

C

I next week same time we are hosting the captain, so bring your questions when they are also very much focused on how to integrate with captain I think today, we've seen great word from Christian, so that will be an inspiration, I guess for all of us and yeah, which one is also again next week and.

A

You're gonna I think I saw you were posting the link to the to the survey here in the chat it's great and we will also post it on in the select channel I. Think right after this one here and once we're done here. Awesome.

A

Thank you and well I hope to see all of you that were online next time and Christian I will I'm pretty sure. We'll definitely see more of you because you actually make here. Thank you so much for your contribution for the time for the inspiration. As you can see, it you're.

B

A

B

Long as we have to stay at home, I think I have a lot of time.

A

That's good! Well, it's great that you're using it the time that you have with us. We really appreciate that. I, don't forget your wife, okay! She she deserves some time. Yeah.

C

You're right, yeah and.

A

Thank you. Thank you.

C