keptn Community, 29 Oct 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keptn Community Meeting - Ocobter 29th, 2020

Description

October 29, 2020
Diary - what happened last week?
Continued working on shipyard-controller
Keptn Community Meetings now included in official CNCF calendar: https://www.cncf.io/calendar/
Hacktoberfest - last week - submit your contributions!
Welcome new contributors - please start with a good first issue to make yourself familiar with the project!
Are you speaking about Keptn? - let us know and get some swag!
Noteworthy Issues & PRs (Review)
For each PR, please tell us if you updated the docs, or in case it was not necessary please say so.
Christian K.
Migrating travis-ci.org to travis-ci.com for keptn, keptn-contrib and keptn-sandbox; Adapt your branch protection settings if applicable
CLA Bot is currently not working - we can not merge anything because of that - we are working on it
Ermin
https://github.com/keptn/keptn/pull/2604 - allow to invalidate evaluation results from bridge
https://github.com/keptn/keptn/pull/2606 - Show all core use-cases on integrations page
Johannes:
(ready for review) Explain requirements for execution plane services: https://github.com/keptn/keptn.github.io/pull/665
(in progress) Wireframes for sequence screen in Bridge: https://github.com/keptn/keptn/issues/2608
Florian:
Invalidate evaluations: https://github.com/keptn/keptn/issues/2449
Shipyard-controller: Built-in task sequence for evaluations: https://github.com/keptn/keptn/issues/2529
Roadmap → KEP / SPEC / Issues
KEP 06: The next generation of Shipyard: https://github.com/keptn/enhancement-proposals/pull/6
Work in progress:
https://github.com/keptn/keptn/projects/29
Backlog Refinement / Next Sprint:
https://github.com/issues?q=is%3Aopen+is%3Aissue+user%3Akeptn+user%3Akeptn-contrib+label%3A%22next-sprint%22
Planning Board: https://github.com/keptn/keptn/projects/29

Learn more: https://keptn.sh
Get started with tutorials: https://tutorials.keptn.sh
Join us in Slack: https://slack.keptn.sh
Star us on Github: https://github.com/keptn/keptn
Follow us on Twitter: https://twitter.com/keptnProject

A

Okay, hi and welcome everyone uh to this captain community meeting it's october 29th uh and uh I see we are a big crowd today- uh welcome everyone. uh Thanks for joining, uh we have a couple of items today and uh yeah. Let's dive into this, um as you might know, captain is now listed also in the official cncf calendar since we're on a cncf sandbox project.

A

So if you have not subscribed to the cncf calendar, you can do so cncf.io calendar or you can still join us by the the usual way via the captain community site, which is hosted on github.

A

Then it's the last week of hacktoberfest, if you still want to contribute to open source, um feel free to do so uh to the captain project or also to other projects. um We have a couple of issues listed with um hacktoberfest with the hectorfest tag. uh Last time I checked uh it was 13 different issues from if you're familiar with golang or angular. I think we have a lot of interesting issues for you, so please consider contributing to captain and uh yeah we're.

A

Also welcoming a couple of new contributors this week, thanks everyone for contributing and uh one part or one item I want to highlight for this week, if you're talking about captain, uh if you're speaking publicly about captain, uh please let us know we might surprise you with some some swag. So if you're speaking publicly about captain on a meetup or on the conference, uh please let us know uh and uh yeah with that, I'm handing over because uh there's someone on my door well and uh I'm handing over to christian.

A

B

If you can unshare, or can I yeah there, we go. Thank you.

B

So what I want to talk to you about is basically something that just happens today, but it's definitely noteworthy for you guys to know um we're seeing a travisci.org backlog, that's building up and that's delaying our builds for pull requests and not just our builds everyone's built, essentially, as travisci.org is being changed into travisci.com or merged actually into travisci.com and they've just started to dedicate more resources into travisci.com, which essentially should help people to give the right motivation to migrate, and it definitely did for us.

B

So I started working on that for captain captain contrib at the captain, sandbox organizations on github and that's just the status update, I'm still working on captain and captain contrib so far, and usually not much should change for your repository. If you are within those three organizations and not that much should also change for your pull requests. The only thing that might change is, if you have a branch protection setup in github, which I will show you what you need to change in in the short demo.

B

So here I have cuban needles. Utils in the captain repository and I've migrated this over from travisci.org to.com just half an hour ago, and I tested this with a pull request here and, as you can see, in the pull request, there are some checks and, as you can see, the checks are now called travis. Ci branch and travis ci pull request.

B

They used to be called a little bit different, which is also something that should indicate to you that you have been migrated. They used to be called continuous integration, slash, travis, ci, slash pr or slash push depending on your configuration.

B

So if you see this string or this label, you are on the dot, org variant, that's to be removed, and if you see the the new one, the without the slashes but the nicer looking label this one, then you are on travisci.com using a github app. So this is how you can spot it on the pull request and if you have a branch protection setup. Basically, you go to your github repository go to settings. You might be asked to log in at some point again or to verify that this is you with your password?

B

That's fine and you might have a branch protection rule, for instance, for your master branch or main branch. If you created the repository recently, then you just need to edit it, and you usually have this. One here require status checks to pass before merging, and if you are on the old version, you might have a checkbox here for travis ci, the old one and what you need to do. Is you remove this check and you add these two checks?

B

Basically, and you save your changes, that's it and then you should be migrated just be aware, when you do that, I believe your old pull requests might break so, for instance, this one by any is now not mergeable, because I basically forced the other two um yeah. I forced these two to run and I'm not sure if they will be automatically run. I guess they will not.

B

So what I will do is, I will revert what I did afterwards after uh this meeting and I'll take a look at the pr and if we can merge it or not, but that might also happen to you and if that happens to you, you can either push a new commit to your pull request or you need to talk to us and we will just figure out what we need to change all right. Any questions regarding this change.

B

Okay, so um the last thing on my side that I want to talk to you is the cla bot on the captain repository it's right now, not functioning, so it blocks our pull requests.

B

This is a known issue right now we're working on it to getting it fixed. Just hang tight, but I hope we can get it fixed by tomorrow or we remove it temporarily. We will see so we can get the pull requests merged and with that I'm handing over to amin.

C

Thank you very much christian. um I'm going to present two poor requests today uh in the bridge. The first one is allowed to invalidate evaluation results from the bridge and uh yeah already last week or two weeks ago. I guess uh this was presented in the core, so we know um supports to clean up quality gate evaluation results.

C

So when you have some evaluation results, those are used to be compared in future, so the results are used to be compared depending on your comparison strategy and also you see the results in the bridge in the heat map. So you can have a look at the previous results and compare them and yeah. Now we introduced the support to invalidate uh evaluation results uh in the core, and now you also can do that from the bridge.

C

So when you select one evaluation result in the heat map, you see the details of this evaluation result uh below and you have a button here ignore for comparison, uh which will open a model and allow you to provide a reason which is optional and yeah. Ask for confirmation, and once you press on invalidate it will remove that evaluation from the heat map and also won't use those evaluation results in the comparison case.

C

Yeah short demo, when I click on my services and, for example, have a look at yeah one of my evaluation results here.

C

As you can see, I have here several evaluations and my last one was even compared with the last two evaluations and, for example, this one uh is an evolution, I'm not happy with, so I can ignore it uh by clicking here and, as you have seen, it's already removed from the heat map, and it also won't be used in future. So also the lighthouse service considers that and it won't be used anymore.

B

A quick question: are you refetching the um the evaluation done, events here or.

C

Are you just removing.

B

It from the local store.

C

No, in this case, I'm removing it from the local store uh on a successful. uh So I I push a evaluation invalidated um event uh to the api uh uh where I post and on success. I remove it from the local data history.

C

So in the heat map here and of course, if I refresh my bridge, then it's reloading those data and also again considering so all the invalidated um results are uh filtered here.

B

Do you know out of your head? What's the limit of results that we show in this chart, 15.

C

50, the limit is 50 results, so maximum, yes, 50, am.

B

So if we were to delete, let's say 30 results because uh yeah we saw that it's missing. We would have to refresh eventually um this. This might be a follow-up uh task, johannes, that we should consider yeah refreshing so that we have the last.

C

D

That yeah evaluations come back yeah.

C

Yeah, actually, if I I have here a one, two, three four, five, six, seven eight results and if I would delete all those eight, I would have an empty heat map here. Although.

E

C

Have more results in the uh data store, for example,.

F

Two questions the first one is: can you please click on the event which you just um invalidated on the sr context? So how does this look in the event stream? Oh.

C

Yeah, I can have the you you thinking about the root event here. I think it was the third one and actually oh yeah good one.

C

The heat map is now, of course, not loading because yeah, I don't have that event here.

C

F

C

I have the evaluation invalidated event here in the event stream, so because it's a new event uh yeah, it pop-ups it pops up at the end here, uh but it's uh of course, uh in this staging uh stage and uh yeah. This actually is uh used to filter that invalidated event, but yeah. The question is: maybe we should just show here the message that this is invalidated.

F

And I would now have one further question for my understanding. Maybe I semester that, can you please go back to the to the sr captain context where you have invalidated one of those.

C

F

um So now I can see in the heat map that the current emulation was compared to this last one.

C

Because the one uh that it was compared to was deleted, so it was actually it was compared to three. uh As you can see, it was compared to three past evaluations, but the one that I invalidated is now also filtered from the heat map.

F

C

F

C

Is persisted so the result is uh not changed after invalidating it. It's just for future evaluations. We don't consider those invalidated results.

F

So my question: um does it even make sense.

F

C

F

And to invalidate own old evaluation results should we only allow to invalidate the last one which was not yet used for comparison? Comparison.

B

So so andreas, I think this question is, is more related to the original issue, where we have a huge discussion about this. um The implementation as it is right now is according to what we agreed on in the issue. I do get your point, although, but we should discuss this in the issue, I think or yeah.

C

Maybe you can focus.

B

We had a lot of discussions about this already, where we thought like yeah, we would need to re-evaluate and re-evaluation, is, is not that easy, so we just started with, let's say, removing or disabling some of them and yeah. Maybe.

C

A side note, the main use case for invalidating uh results is for your first or let's say the initial setup, where you don't have your slis configured, for example, or for failed results that you just don't want to appear anymore in the heat map.

C

That's the main use case, so you invalidate, for example, field results and field results are usually also not used for comparison.

C

Of course, you can also invalidate successful results, but I think that will not be the main use case.

F

Thanks for the explanations here, so at least um one follow-up issue I definitely see is and that, if you cannot load the emulation results because it was invalidated, we somehow have to say the user yeah. This was invalidated.

F

It's expected behavior.

C

Of course, yeah nice, thank you for the feedback and uh yeah one more pull request. I'm going to present today is show all core use cases in the integrations page.

C

So uh I think two or three weeks ago we presented the integrations page and on the integrations page, we show um some examples for the use cases, or actually we had just one use case there: how to trigger a quality gate evaluation and now we also add the examples for how to trigger deployment with a new artifact and how to trigger remediation with a problem event for the cli and for the api.

C

This information is only shown um if you use captain uh not in the quality gates, only installation, so in a quality gates, only installation, you will see only the trigger quality gate, evaluation examples and for the full installation. You see also the other examples which are in this case, as already said, how to trigger deployment so for the cli.

C

It's the captain cent uh event uh new artifact with the uh placeholders for project service, image and tech also for trigger remediation each it's a captain sent event with the problem in a json format and in the api case yeah. It's simply a post to the event endpoint with having either the configuration change event or the problem open event uh in the payload um yeah, that's all from my site. So if there are no more questions, I'm handing over to johannes.

D

I mean just one uh quick side note uh before or before reviewing your pr or closing it. um There is a follow-up issue, especially in regard to your to the environment, variable that you're using to control whether we showed this use case or that use case, and um I would like to discuss this follow-up issue or actually, I will show it in a couple of minutes, um as I would propose to to set the value for this um environment variable um depending.

E

D

Wire helmet shot first, and also when installing captain yeah and um also but uh oh let's, let's do it uh later on to discuss it: okay, yeah! Well, then it's my turn.

D

um What I have to show you is actually that I continued working on my documentation issue that explains the requirements for the captain services that are running in the in the execution plane. um Yeah.

D

Last week we had a um a review here in this community meeting, where we took a look at the implementation or of the of the improvements that I did, and I also then uh took your recommendations and improved my current pr and when we now go to custom integrations and to the section of writing a captain service, then I'm now showing you two options for your distributor.

D

One option is the way that you run your captain service inside the camden control plane. Then you get now here a table that shows you all the settings you have to configure for your distributor to to make it work in a control plane, and in case you run it outside of captain your captain service um yeah in the execution plane that is, can be somewhere, and for this I'm also presenting you now a table that explains all the different settings you can configure, depending on what you want to do all right yeah.

D

This was then uh improved based on the feedback I got last week and I'm now at the state where I would like to ask someone for for reviewing my pr.

D

I mean when someone has time the next couple of days. Please just give me a review on that would be really appreciated, and the second issue I was working on this week is a wireframe for the sequence screen in the bridge.

D

Well, I was doing this in jointly with with airmen, since we are now changing quite a lot when it comes to the eventing mechanism of captain uh yeah, especially in with the new shipyard controller. We also need to rethink how we present this in the ui, and for that we came up with the idea of providing a new screen, which is then called the sequence screen right now.

D

It's in progress, this wireframe, but by next week on thursday, we are ready to show you um a preview and also how it should how it will look like, and then we also want to discuss it with you in this meeting, all right, yeah. It's currently in progress, but already looks good and I'm really excited. I'm really excited about the new screen, as it will help a lot.

D

All right that was it from my end, then I will hand over to florian all right. Thank you.

F

G

Okay, let's get started today, I wanted to show you two pull requests, but the first one actually was the implementation in the backend part of captain for supporting the invalidation of in evaluations.

G

So I think, uh since I mean already gave us a great demo about that feature, I think we can skip that one and just continue with the second one, which affects the shipyard controller. So let's just have a look at the issue itself. So basically, the requirement here was that the shipyard controller subscribes to trigger events that are defined in the shipyard yaml and provides a built-in task sequence for evaluations.

G

So in order to know what that means, let's just have a look at an example of the shipyard. So as you, we have already seen this new format of the shipyard a couple of times.

G

But what was meant in this issue here is that, as you can see here, each of the stages provides or has a couple of sequences that can be triggered, and that consists of several steps that should then be executed by the execution, plane services and in the shipyard controller, it's now possible to listen for an event that has the name of the correlating task sequence and then trigger the sequence and start the complete task: sequence flow and synchronize all the incoming finished and started events in order to to progress through that.

G

So, for example, if you would like to start the the artifact delivery sequence in the stage dev, you can now send an event with the name.

G

Sh.Captain.Event.Dev.Artifactdelivery.Triggered and also pass some input data that should be sent to the first service. That's part of this sequence and the shipyard controller will correctly pick up this event and start the correct sequence.

G

That was the first part of the issue that has been implemented and the other one is, as you can see here, we do not have any sequence, that's been uh with the name evaluation and that means uh yeah. And basically, if you would like to use the qualitygate standalone use case, we added the convenience feature in the shipyard controller that allows you to use that use case without defining a separate sequence for that.

G

So now, if you, for example, would like to start an evaluation in the stagedev for a certain service, you can still send an event with the name: def.event, dot, evaluation.triggered and then the shipyard controller will look at the available sequences and, if no sequence with the name evaluation is present, it will still support that and send out an evaluation, the triggered event, and that in turn, will get picked up by the lighthouse service and then the whole evaluation. Workflow will start.

G

And, of course, you can always override that. So if you would like to define a custom task sequence for an evaluation, you can do that by adding a sequence with that name and then starting it with the evaluation.triggered event.

F

All right, I would have two questions um to those two points. My first question is: is it not possible to do an evaluation for a continuous delivery use case so um before there was no possibility to to trigger an emulation without um deploying an image into the next stage, and is this use case now possible.

G

Yes, so, for example, if you would remove that step here, this deployment step, you could still it would directly proceed with the test where it will test what is currently deployed, and then the with the.

F

Evaluation, so does this mean I can use the artifact delivery sequence and also the emulation sequence um for the same project service and stage that would be possible yeah, okay, cool. This was my first question and the second one is now um we are calling this artifact delivery, the sequence name, yeah, artifact delivery and before we always called it um configuration change, so we somehow need to streamline this.

F

Mr my help service, so the helm service expects in the in its event, in the triggered event, some some input data for the configuration change either we call this configuration change or artifact delivery.

F

This needs to be defined. I guess, but.

B

F

Can discuss this on the issue.

B

Yeah, I think the labels or the strings that we use are not matching so helm service uh wants configuration change somewhere in the payload. If I'm not mistaken, and we call it artifact delivery here. um Yeah should be a minor change. The name.

G

Of the sequence shouldn't matter, but the helm service will listen for the deployment dot triggered events, but still here we are not consistent, since we are providing the configuration change property directly in the data block of the event, and in my opinion it should be part of the deployment property. So within data dot deployment, dot configuration change. I think that would be more streamlined in terms of naming.

B

Do we even need configuration change? I mean what would be the cli call or the api call so to say that we would use with that. We would send in. We would trigger an artifact delivery and the parameter of artifact delivery would be deployment and deployment would have. I don't know an image, or maybe a url, to something depending on what your deployment services.

B

What or did we specify this strictly.

G

In that case, we're now more generic. So now we do have that configuration change property which is actually a map that maps a strength to an interface type and that allows us to provide arbitrary values to be set within the helm chart.

G

And previously we had a dedicated field called image and attack, and that was a little bit too limiting, because using that we, for example, didn't have the possibility to change the value of the replica account, for example, or any other property within the helm chart and using this generic mechanism. We we can do that now,.

B

All right yeah, then, let's create an issue and compare the the variants that we just talked about. Johannes, can you do that or should I do that.

B

If you want, I can do it if you, if you do the next part in the meeting for until.

D

G

All right, uh if there are no further questions, those were the two pull requests. I wanted to show you, but also really quickly. uh There were also some bugs fixed uh in this sprint, which I'm just going to uh show you which they were, and so that in case you were following those you now know that they have been fixed, so really quick, the first one when the cli didn't have a connection to the captain cluster.

G

It ran into a timeout, even though, when, for example, for some command, it wasn't actually required to have a connection to the cluster, so, for example, for the help command- and this has been solved in this issue. So now it won't hang for 30 seconds anymore, but it will return after a predefined timeout of 5 seconds all right. A similar issue came up in the helm service when it was waiting for a deployment to be ready.

G

So here we also didn't have a fixed timeout threshold and that caused the service to hang for an indefinite amount of time, and that has been also fixed here so now after uh I think three minutes. If the deployment isn't ready, uh the home service will now finish with an error and finally, the last one.

G

So the way this bug came up was it was during the migration or during the upgrade from a previous captain version.

G

For a project where an upstream repository has been updated after the project has been created, so with the captain update project command, it wasn't visible in the bridge that the upstream repository actually was configured and the root cause for that problem was not in the bridge, but in the configuration service, because this update hasn't been picked up correctly.

G

But now this has been fixed. And now, if you do the update project, the upstream repository will correctly show up in the configuration service.

G

All right- and I think.

D

I'm now, through with all my no uh florian, there's, there's an additional one, the event endpoint.

G

Oh yeah you're totally correct, uh so basically here we, uh I can quickly show you that one.

G

So here we just specified the wrong content type header for the responses of the mongodb data store. So here we had producers, application cloud events plus json and application json, and in fact this was not correct here. So it only accepts cloud events and json, but it doesn't produce that all right.

F

I think that was it.

F

Why is it not the cloud event plus json, what we return from them? Will it be.

G

uh Because it doesn't return cloud events, it returns an array containing cloud events, but this array.

B

G

Already is not a cloud event anymore,.

B

So the full story is a little bit more complicated than it's on the issue, but essentially nginx blocks the response and just tells like I don't know, http 406 or something like that, which is weird because if you contact the service directly, it works. So we just went the easy way. But we have a follow-up issue that we need to do some research and see what's actually happening or if, if there's something wrong with nginx or if it's maybe a h2db spec.

D

Topic all right, yeah, then, let's jump over to the working items that are ready for review. I will skip those that are not closed yet and are waiting for for a re. Final review here. Burned also worked this week on doing an update on the documentation this one. I can actually show you because what he did is he updated the part where you can set up monitoring for dynadress and the install section of dynadress.

D

He updated this screenshot for the api token, since dynadress has now two versions of the api, v1 and v2, and you have them to toggle the proper feature flag in depending on what you, what you want to need or what you want to use and he updated those with slightly changes here down on in the text.

D

I think we'll move it over and I opened an issue thanks to uh christian heckleman, um who brought it up that the help that the help for the use case flag for a captain install command is not very useful before it said the use case to install captain for continuous delivery and well.

D

That does not help that much and we improved it by now saying when selecting this flag or use this flag to install the execution plane from continuous delivery without this flag, your captain is, is capable of the quality gates use case and the audit automated remediation use case, and this was then implemented by nanik thanks a lot and um yeah with captain auto date. We you will then get this.

D

D

D

Then, let's go over to those that are in progress.

D

Christian already told us that he is currently in prague in in in migrating the repos from travis.org to travis.com. It's a critical one, as we are kind of blocked with all the builds in the background. They are cute and we don't get any any successful or it takes a long time to get the bills done.

B

Yeah, the the bad part actually is, as the issue says, by december 2020, so we had this in our backlog. For I don't know three four weeks yeah, it got critical today.

D

um Kristen how's it going with the documentation for the troubleshooting section.

B

Yeah I was sidetracked. There is a draft pr for this, but I still need to refine on it.

D

Then florian, what's your progress on updating the dynadress service so that the dendrite service understands the new cloud event.

G

Almost done, but two follow-up issues came out of it, but I think that task can be uh will be ready for review tomorrow.

D

Then here we have the issue that says: adapt, cli command, create service, onboard service and delete service to use the endpoint of the shipyard controller.

G

D

G

Same here almost done, but in order to merge that we have to wait until the the previous task for where I've implemented the actual creation of the of a project within the shipyard controller to be merged. So today we had some problems with travis, so it will be merged, I guess by tomorrow, and then that pr will be ready as.

D

Well perfect: this means we are kind of depending on that one.

G

D

All right, then, we have the issue and exit fire. The captain's bridge, airmen how's it going.

C

uh Yeah, we have actually those two refactoring tasks in the bridge that are still open uh yeah. I had no time this week uh to look at them, but I hope I will have time tomorrow uh to yeah finish them.

D

All right yeah then, as I already said, uh yemen and I we are working on wireframes. uh We will present them next week.

D

This one is on the review and then here we have one um that is that is um implemented or going to be implemented by tite. He has opened a pull request and also, um I think, florian left a comment there.

D

Let me just think tide is also on the call. Let me promote him to panelists really.

D

D

Hi tide: um do you need some help with your issue or are you okay? So far,.

E

um You already made a comment on what to do and I think he it he did it already. So I just went through it and I I reviewed the yeah the forward and the receiver yeah. So I think I got I go what to do there? Oh, that should rather yeah immediate comment.

D

All right, yep um florent the review and left a couple of comments, but please let us know when you need help.

D

All right and then here further down, I have a couple of issues that are also in progress and um people from the community are working on that one. On those. um We also left a couple of comments there to get them finished.

D

Yeah, let's keep it there and let's wait until these colleagues are done.

D

Then we skip the assigned and committed column and we'll jump over to the backlog.

B

Yeah, so uh maybe I'll take the first one. This is the follow-up issue to what ben did in the docs. um We already have the the issue linked up there and there should be a pull request linked with the issue. I guess so. What we need to do is to update the tutorials with the new screenshot. That's linked in a pull request.

B

um I'm I'm not fully sure might be two or three parts in the tutorials where this needs to be updated. So yeah should be rather simple because we can reuse the screenshot and just make sure the texts still make sense.

D

I think uh really straightforward, as we have already the screenshots and um yeah.

D

Someone up for doing that. One.

B

Oh, we can leave it in a backlog. I guess it's a good first one. So if we find time we'll just do it.

D

All right, then, the second issue that was identified by jurgen last week, where he observed the following behavior.

D

The problem is that this time- or it is this duration does not change, um does not change and will always be the same depending on the evaluation you're. Looking at here, for example, um airmin, ah nothing I mean jurgen selected is taking a look at the last evaluation. This had a duration of five minutes and when he goes or click through the heat map, the duration never changes.

D

It's it's always the same, but they were different as we.

E

Identified yep, okay,.

D

um Is someone who is someone has someone time to work on that issue?.

C

C

E

Is it? Is he the first um good first consecutive, the first issue.

C

um I don't know.

D

What's your recommendation.

C

It might be a small issue just because uh yeah, probably the duration is already calculated somewhere, but just not updated when clicking on the element in the heatmap. So.

C

Yeah, it could be a good first or easy to start.

D

And ty: do you want to give it a try, or would you like to focus on the other one? First.

E

Yeah, I think I would I could I could give it a try. Yeah, okay,.

D

All right thanks a lot, and I will assign it to you or actually, you have to leave a comment on this issue that I can then assign it to you just uh say that you want to work on that. One.

E

Yeah so yeah wrong, evaluation time: okay, yeah.

D

All right, then, the next one is the follow-up issue that I was mentioning before, uh because yemen told us that the bridge now has the environment variable. That controls, which use cases are shown in the integrations page and this environment variable is called captain install type and here a reference words set, and I would now propose that this variable supports three different types, which are the quality gates. Then continuous operations and continuous delivery.

D

And when we do a captain install- and we just do a captain installed, then we set this in very environment variable to continuous operations. Because what we installed, then, is the quality gates and the remediation use case.

D

And when we do a captain install with the flag minus miners use case, then we set this variable to continuous delivery.

D

In order to provide, then the full full feature set in the bridge.

F

I would have a question so I did not treat this issue beforehand so should we really solve this using an environment variable or should we extend our api or metadata api which allows you then, to query this information, so is that the canon installation itself knows, but it's it's a quality gate, emulation, continuous operations or continuous delivery operation or.

D

But I mean I understand your point.

F

You so we don't have to decide it now.

B

Also so uh just to weigh in on this with my two cents here um right now, we just need to have the distinction. What was installed with 0.7, whether it's quality gates, continuous operation or delivery in the future, we will need a service list that has been installed and maybe, depending on that service list, we will show different output in bridge or whatever. um So this will change in the future. So uh we shouldn't over, complicate it right now, but I definitely get your point.

B

It would be better suited in our metadata endpoint um uh long term, but I I guess for now it's the way it is with the helm chart and as an environment variable works for us, I guess um maybe long-term johannes. We should figure out what metadata we actually need out of a captain installation, because it's.

E

B

Just the distinction between quality gates and delivery and whatever, but there might be different ones.

D

D

Okay, but um um apart from the from the implementation on the api side, does this make sense to you to control the bridge? Why are they environment variable the test and the value did one of these three values.

F

I mean it's a technical solution. Yes, so uh for me it's not quite clear that we return continuous operations when we only say get an install, so captain quality gates is then not.

D

That is true, because uh continuous operation kind of includes the quality gates use case, as it is part of the remediation use.

F

D

Or let's, let's phrase it in a different way: the quality gates only is the smallest possible set of services that you can use with. Captain then comes the continuous delivery on top.

F

But you can quality gates only now, that's the point.

D

That's the point yeah and that's uh that's the point, because when you do a captain install you install quality gates as well as the remediation part.

F

D

F

But um cool for this issue, so that's.

D

Fine all right is it then clear to everyone, because then I would like to resign it to someone.

G

uh You can assign it to me.

D

Cool. Thank you very much.

D

Oops, I clicked on my name, sorry for.

D

D

All right, yeah, then the next one is about the gatekeeper as now, more and more functionality moves from the gatekeeper into the shipyard controller, there's not that much left over from the gatekeeper, and it should become kind of just the approval service. Who is then taking over the responsibility of approve of of accepting or declining an approval, and this is what is written down here.

D

um It's quite a lengthy issue or issue description, but the overall goal of this issue is that the gatekeeper is then is just capable of doing then, the approval step by listening to an approval, triggered and then sending out an approval started as well as an approval, finished.

D

D

D

It's some has someone time to work on that one throughout the next week.

G

D

D

D

All right, then, I have, I think, one left.

D

Which is about linking the stages um yeah linking the stages, based on on the events that are sent out by our task sequence? Just let me give you an example here or in this new shipyard.

D

You see that there are two stages: a dev stage and a hardening stage, and both have an artifact delivery sequence and whenever or when this sequence is finished, it should kick off the execution of the sequence in the hardening stage, as this is here configured by the trigger, which is def dot, artifact delivery, dot finished, and this should then link the stages you have configured in your shipyard according to florian. This is actually working right now on the master.

D

But what we want to address with this issue is the is the update of the integration tests for um for testing this out, and we left a note here, because then you wanted to provide a description of the integration tests that you would like to see.

B

Yeah I I guess I forgot that um I'll. Add that on my to-do list, I'll add the note by tomorrow.

D

Yeah, I think we we can also leave this one in the backlog and when the descriptions are there- uh and someone has time to work on this issue then feel free to take it and and to start working on writing the tests.

B

Yeah, maybe just to summarize it quickly. We have integration tests that deploy to google kubernetes engine and do an end-to-end test, and I want to avoid it for such a basic feature that we need to rely on this end-to-end test. It would be much nicer if we just have now that we have the ability to control the shipyard content to have an integration test with two stages in the shipyard and the custom workflow or whatever.

B

That's like two flies with one clap and we have it tested plus. It would also work for pull requests because it could be tested locally, using keys um and you don't need a cloud environment for it.

D

All right, yeah and then like every week. I want to bring up this issue.

D

It's just a summary of all those issues that are out in the captain contributing sandbox organizations, because we need to upgrade or update other services in order to make them work with the new shipyard controller and take a look at this issue and feel free to grab one of those further down, as they just tell you. What need to be upgraded to make the captain contribute sandbox services, aware of the new shipyard format, no, no new cloud events, format.

D

All right, yeah, that's from my end, um and I don't have any further item on my bullet list. Except of that one. When you want to start contributing to captain, just feel free to take a look at the issues that are marked with good first issues or with hacktoberfest that are also issues that are good, getting started, issues and leave a note. There just inform us via slack that you want to start working on an issue and then we will assign it to you.

D

All right, we have a couple of minutes left. Are there any questions from the.

E

E

So I would like to ask this question: is there like um a detailed pitch where I could get the overall architecture of what captain does like the old design systems yeah.

D

The best that I can provide to you is shown here you go to.

D

Captain.Sh then under concepts there, you find an explanation of all the use cases that are supported, which is the multi-stage delivery, the quality gates, the continuous performance validation and the automated operations use case. And then here you find an overview of the architecture captain is built on.

D

This is okay, to some extent, a very technical one. Here you see all the components and how they are kind of working together.

D

It's the entry point is an api gateway which is running on nginx, and then this one is speaking with the api service and uh yeah in the center of captain. There is, then, um nuts for doing the eventing and, based on that, we have done implemented a bunch of services that are listening to events and do their job. Depending on on the use case you have installed.

D

I think it's a good starting point to get an understanding how everything ties together and works together. You also find short explanations of those components further down.

D

I would recommend you just a reading over this documentation and let us know when you are missing something, because we are always looking for improvements also for in regard to the documentation and when you miss some information. Just let us know please.

E

Okay, thank you.

F

But this is a really good point, because we need to update especially this site for 0.8. So, for example, we don't have any event broken anymore and I would propose them to create the issue for those and also label it with zero point.

D

Zero point: data yeah because the the shipyard service is gone, yeah.

B

Also uh regarding the issue, I've written my text in the link stages based on task sequence issue, and, while writing that I just figured out for such a simple integration test, it might make sense to have a so-called echo service somewhere, even either in captain sandbox or in captain core. I don't mind. um The echo service would really just get a triggered event reply with a started event, then five seconds after that with a finished event, so it doesn't do any logic. It's basically an empty service that really just echoes what it gets.

B

That would make integration tests really really simple, and I think the service itself is is not too hard to write.

B

Yeah just putting it out there, I guess we can discuss it on the issue because we are running out of time now.

D

That's true yeah. We are on the top of the hour when there are no further urgent questions, just waiting a couple of seconds.

D

No. There aren't then um thanks for joining today's community meeting and um have a nice day and see you next week. Bye.

C

C