keptn Community, 5 Nov 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keptn Community Meeting - November 5th, 2020

Description

November 5, 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/
New webinars coming up:
LitmusChaos testing + Keptn: https://dynatrace.zoom.us/webinar/register/WN_2j3CtVvcR1iMqeRtoE7X8g
Improving SLOs with Prometheus + Keptn: https://www.youtube.com/watch?v=f9tRg2DWh48
Keptn in Gitlabs #everyonecancontribute café next week: https://gitlab.com/everyonecancontribute/general/-/issues/56
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:
DockerHub / Image Pull Back Off https://github.com/keptn/keptn/issues/2294
Pull Through Cache: https://docs.docker.com/registry/recipes/mirror/
Gabriel Prioli:
initial release of the Ansible Tower service: https://github.com/keptn-sandbox/ansibletower-service
Ermin
https://github.com/keptn/keptn/pull/2604 - allow to invalidate evaluation results from bridge
Florian:
Added environment variable with use cases: https://github.com/keptn/keptn/issues/2609
Use shipyard controller for creating projects/services: https://github.com/keptn/keptn/issues/2557
https://github.com/keptn/keptn/issues/2266
Delivery assistant for Keptn 0.8.0: https://github.com/keptn/keptn/issues/2533
Distributor: https://github.com/keptn/keptn/issues/2562
Update CloudEvents in Dynatrace-service: https://github.com/keptn-contrib/dynatrace-service/issues/181
Johannes:
(ready for review) Explain requirements for execution plane services: https://github.com/keptn/keptn.github.io/pull/665
Wireframes for sequence screen in Bridge: https://github.com/keptn/keptn/issues/2608
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

Hi and uh welcome.

B

A

uh To this captain community meeting uh it's november 5th, um we don't know who is the next president of the united states yet, but we have a lot of great topics for this captain community meeting and let's just jump right in uh so we don't know how you found the community meeting and how you joined.

A

uh Maybe it was via the community meeting calendar on captain.sh and on our community site on github or via the cncf calendar, um just as a short mention captain community meetings are now also listed in the official cncf calendar so make sure to subscribe to this calendar. Also, all the other cncf projects are listed there and you might find some other interesting community meetings or sick meetings uh make sure to subscribe to this calendar to get the next captain community meetings also directly into your calendar.

A

We have a couple of interesting webinars coming up next week on wednesday. Next week we have a joint webinar together with the litmus team. Litmus. Chaos is also a cncf sandbox project and we have been working together with them on an integration between litmus chaos, testing and captain, and the integration is really about testing and evaluating the resiliency of microservice applications uh with litmus chaos and captain. We presented this already, a short abstract or short excerpt of the the webinar we presented yesterday in the app delivery sick meeting of the cncf.

A

um The demo was very well received and in the webinar next week we will go a little bit more into detail and we'll show you how you can have chaos testing orchestrated by captain directly in your cd, and we will explain what are the benefits for this integration and on tuesday next week. Actually, I should change the order here, but anyway we will have a joint webinar with um julius faults.

A

uh He is the co-founder of promethos and he's the founder of his new company called prom labs, uh and we will talk about how to improve um slis and slos with prometheus and captain. So we will see how, with the help of a of an open source tool, um we can write better uh queries, better prom, ql queries, improvement.

A

First, to write better slis and better slos and to include them in captain, so this will be on on tuesday we will stream it directly to youtube, make sure to subscribe to our channel, and um it's already like there is a placeholder already up and running, and you can set yourself a reminder that you don't forget to to join our webinar.

A

um Also next week um captain will be presented in the kid labs. Everyone can contribute cafe. um There is one of our captain community members. I think he is also joining today. uh Christian heckleman, hi christian um and I know you're talking hi.

A

uh Do you want to say a word uh for um uh what it actually is.

C

Yeah, that's one thing I want to to know by myself, but as far as I understand it's a more kind of an hands-on a session where we are or where I will show how we have integrated captain into into hitler pipeline. Therefore, I have um also created an on on the gitlab sas environment.

C

So it's reachable for everybody, a demo integration for for captain and also published our internal, not internal, but our uh captain templates for gitlab there. So everybody can reuse uh the captain integration for gitlab and we will go through uh yeah how to deploy service and or or utilizing captain within a gitlab cicd pipeline.

D

Is that something that gitlab per se is organizing or are.

C

D

Hosted by gitlab.

C

It's I I just know it's it's and gitlab event.

B

C

From um what is this twitter handle being esme, the michael, oh, what's his name.

D

Okay, yeah, it's okay! I was just wondering if that's really hosted by github or if it's just gitlab.

C

Get gitlab sorry.

D

C

Yeah, it's hosted by gitlab and alois actually pulled me in to this. So.

A

Cool yeah, thanks christian for for doing this uh and spreading the word um around captain. That's really exciting. Thank you, um yeah. uh If there are for the for the broader captain community and for for those who have also joined this meeting, uh if you were talking about uh captain and speaking publicly about captain or blogging about captain, please let us know uh we have some captain t-shirts that are pretty nice and uh we want to send you as a gift.

A

If you are um speaking uh publicly about captain actually, today, I'm not wearing a captain t-shirt. uh I think. Maybe you can see my captain hat there, but yeah. We we.

B

Do have some some.

A

Nice captain swag and we're happy to to ship it to you. um So with this, these were like a community updates. uh With this we will jump into the more technical part and we have a lot of technical issues uh that we want to discuss today or typical topic technical topics and I will be handing over to christian and then we will go through the list. So please christian, take the screen. Yeah yeah.

D

Thanks such as.

A

Members, sorry, I would just uh say one more um one more sentence. If you have any questions, uh please let us know in the chat or in the q a and we will be uh happy to address them.

D

All right thanks jung. I just added myself on the top of the list, because I have more of an organizational topic or something that's not so technical but might affect a lot of users um so starting november, first or even second docker hub is introducing and pull rate limit for anonymous users and authenticated users. If you are on the free plan, this pull rate limit is 100, image pulls from docker per six hours.

D

um Usually it shouldn't affect you if you're an anonymous user, but if you're, trying out captain and you've been continuously creating a cluster deleting a cluster, creating it again, etc or you've been generally working a lot with docker images from docker hub or if you are behind the company firewall, and you only have one public ip address, you might run into this limit and this will result as I've written on a public slack already in an image pullback of error.

D

There's nothing that we can do at this point in time to reduce this other than pushing our docker images to a different registry.

D

But what we've done so far is we've applied on docker hubs, open source free tier, whatever it's called, which we will get news from. I hope soon once that's happening, um some of our namespaces, like the captain namespace and the captain, contrib namespace and captain examples. I think they should be covered by this, and this means that, even if you are in the free plan and an anonymous user you, the pull rate limit, does not affect images in the kept namespace. So this is the kept namespace on dockerhub, not in any kubernetes cluster um yeah.

D

I will monitor this. Actually, we will monitor this and, if anything happens, we will notify you um yeah.

C

I have a little question here: it's it's not captain related, but for the rate limit calculation. Is it really on ip address based or wasn't it on? I know there was a discussion outside on on gitlab as well. Isn't it based on the docker engine id from every surfer.

D

uh Interesting question: I cannot answer this from my understanding. If you are an anonymous user, there is no way for them to tell who is who, so, if you're, behind one public fp address.

C

Yeah but but when you're, installing docker on your machine, your docker mach or the instance, a docker demon getting in not getting him but generating an id which he then sends to server and that there was a discussion with gitlab, I'm not really hundred percent sure.

C

um But then it is much not or more less uh affecting these. This issue.

D

Because when you have it's certainly better, if you're behind the company firewall.

C

But not true, but if you're behind the company firewall, it's really easy to up install a pull through cache.

D

Yeah, that's a good point. This is something we can also show people. So docker has a built-in feature registry, as a pull-through cache that you can use, and you need to set it up as a company you can have it set up as a caching server in between that's quite nice, and you need to configure some things in the cache and in your docker local config file.

D

You would have to configure a registry mirror so to say this would only be valid for pulling images not for pushing images, and that would be one of the solutions.

D

I will actually share it in the document as well. One thing that you cannot do, although is you, cannot easily configure this on your kubernetes host, so on your kubernetes host. This is, I don't know where you would put this so if anyone knows feel free to comment on the issue anyway. That's why we're raising this point. So if you see such image pull back of errors, it's most likely not a captain issue, but a docker rate limit issue and there's nothing.

D

We can do other than pushing the images to a different registry or waiting until we get the free tier for open source, and now my zoom is trying. I cannot actually stop sharing uh you already did it's always.

A

D

Yeah, okay, uh did you want to share again or it's? No, I just wanted to stop sharing it's. It's still showing me that I'm sharing, but then it's fine I'll just rejoin the meeting.

A

D

That's it from my side, thanks.

A

Okay, thanks christian uh and uh yeah, if you run into any issues uh installing captain or um operating captain because of this limit, uh please let us know- and we will uh we'll also actively monitor this issue. um So the next on the list is gabriel from the captain community.

A

He has worked on an ansible tower service uh gabriel. Do you want to take the screen sharing hi can.

E

You hear me: yes, all.

A

E

Okay, uh I'll do a quick one, since we have a lot of items in the agenda is basically an update and letting you guys know that we've um we've made an initial release on the ansible tower service. That, um of course, pretty much the beginning, still lots of stuff to do, but you myself and eugene have decided to already push that as a release, so that we can start tracking progress by the issues and, and everyone can start using it as well.

E

So there is, there is documentation here for those that are interested in trying it out. um There are some prerequisites since we are integrating with ansible tower. There is gonna there. An enhanceable tower instance is needed, but then the rest of the installation here is um pretty much kept in service some secrets and uh sample application to onboard it.

E

The result that we wanna we wanna have is something like what we see here in the bridge. uh A problem is opened from a monitoring solution. Then the remediation service kicks in, because we had already defined that this has an a remediation.

E

Then the action is triggered by the remediation service and then as a source. The ansible tower service starts the action and that's the actual job that runs the ansible playbook. The action finishes. um An evaluation starts to see if it was successful in correcting the problem. uh That includes a retrieval of sli that in turn, results in the passing or failing of the evaluation, which is the way that we are defining if the remediation should be concluded.

E

In this case, it was a pass, so remediation is finished, and then, additionally, we receive an event from the uh the monitoring solution that the problem was resolved. So this is the um going back to the concept. This is um concretely the automated operations and uh here's a summary. If you need to explain to someone, this is a good good way, so the problem is alerted same thing that I just said but get gets the remediation action and executes it.

E

Re-Evaluates the slo and then the escalation would be a second point which would be an another action. um But I think that is pretty much it. It's just an announcement um that we are well that this is working. It's it's it's beginning to be in a working state, and I just want to thank jurgen and johannes, for you know following and helping me get through this. uh It's been really really interesting to develop this one.

F

Really cool work thanks.

D

Thanks gabriel, thanks johannes, is this uh helping us towards uh more remediation use cases.

F

Yeah definitely yeah cool and the way that gabriel implemented. It allows us that actually any job template can be executed, and this brings in quite a good flexibility.

D

Nice thanks gabriel awesome.

A

Cool yeah thanks again uh and we are moving on to the next uh item. uh It's airmen's turn to share his screen and talk about his contribution.

G

Yeah, thank you very much. uh I'm going to present one pull request today. Actually you remember that one from last week's community meeting, but we had a discussion about some visualizations and some adoptions here that I want to present you today.

G

So the changes I will do a demo a live demo are when you invalidate an event, and I will for that look up. One um of this older configuration changes where I know that I invalidated the evaluation result.

G

You now see the evaluation done tile here with the additional information that this evaluation result has been invalidated and for this result, for this reason you don't see the heat map, so you don't have a comparison, but you still have the option to show the slo and see the actual result of this evaluation done event, and also the sli breakdown in case that you need it for some reason.

G

Also. What changed here is that, although in the slo file, it says that the number of comparison results is three, it says that this comparison uh or this evaluation uh compared uh with last two past evaluations and that's taken from the evaluation details there, we provide the compared events and, as you can see, there are just two events used for comparison, because uh before or at that time there were just two uh evaluations that passed and it had only two uh those two to compare with.

G

um Also what changed is in case that yeah, you have your.

G

Evaluation result in the heat map now and it was compared with two past evaluations. You see those two also highlighted in the heat map in case that I choose to delete that one or one of those two that were used for comparison comparison.

G

I can yeah of course, delete it or invalidate that one and this evaluation was, of course compared with the last two past evaluations, but one of them is missing now in the heat map, and you get those additional information here. So you see, one result is not showing up in the heat map because it has been either invalidated or it is older um yeah. That's what changed here and so now. I think everything should be um yeah good explained in the bridge, so nothing should be missing here.

G

Are there any questions, or otherwise, I'm heading over to florian.

A

Thanks erin for for sharing this, there was actually a question today in the slack uh I I had a discussion with a community member on how captain is doing a comparison with multiple events and since you're sharing this right. Since you have shared this, uh I think yeah, nice.

H

A

H

All right, then, I continue uh today. I have a lot of ground to cover, so let's just jump right into it. So the first issue I want to show you was a slight extension to the helm chart and the installer.

H

So, as you might have seen, we now have a integrations page in the bridge which basically gives a nice overview of the available use cases that captain supports and how to use them. And of course, captain allows to install itself in various versions. So, basically, depending on the use case you want to to have, you can either go for the full version.

H

That's uh that supports the continuous delivery quality gates and continuous operations, or you might want to just have the limited version where you only use the quality gate and the continuous operations, and this is specified during the installation and therefore I have added a environment variable to the bridge deployment in the helm chart which is set according to the provided input for the install command.

H

So by default, if you know uh or actually that the correct values are here so for the default case, when you install captain just with the quality gates enabled uh and the continuous operation, then the variable will be set to this value, quality, gates and continuous operations.

H

And if you choose to use the continuous delivery use case, then that will be included to the list of available use cases as well. And then the bridge can react to that accordingly and display the proper information.

H

Are there any questions about that?

H

All right? If not, I will go on with the next one or the next two issues. Actually, so those were quite a big change, so, as you might have seen already, we introduce or we're going to introduce a new shipyard format in captain 08, which looks like the following just really quickly.

H

We now have yeah, basically a different format with a different api version, and the task of this issue was to make the shipyard controller or new service that's responsible for handling the shipyard and the task executions within it to be to support the creation of new projects using that specification and also to create new services and delete them. If you choose so, and therefore I've added a new endpoint for those operations for the shipyard controller and as a consequence of that, the endpoint for creating projects and services has been removed from the api service.

H

Because now the cli will use that new endpoint of the shipyard controller to create new projects and also what we have changed is we uh do not use the websocket communication anymore when creating projects or services.

H

But since now the complete all the functionality is contained within one service, which is the shipyard controller and not distributed among multiple services.

H

We now have the possibility to just provide the the result of the project, creation or service creation uh directly within the http response, so that will also make it a lot easier for clients that use the api directly. Instead of the cli in order to know what's going on, and if the project creation actually was successful, all right and the second issue that was related to that was basically the update of the cli, because.

H

No, that's not it.

H

Here it is so also the cli needed to be adapted so now the create service and onboard services delete service commands, use that new endpoint and also the deployment strategy will not be supported anymore, because now we have a way in the shipyard file to to handle different deployment options for different services.

H

All right and also the shipyard files for the integration tests have been updated in that issue as well. All right any questions about that.

H

If not, I will continue with the next one, so that was a collaboration between uh god's tight and I so he was nice enough to take over this issue and provide a pull request, and I just did a few minor modifications on top of that, and actually the issue was that, since now also the api service within the captain core will use the distributor to send event that has received via the event http endpoint, but it doesn't actually listen to events coming from the nuts server.

H

We now have also the the requirement for the distributor to also work if there is no pops up topic or pops up recipient uh defined. So in case, if you want to use it only for forwarding, you shouldn't have to set the values for the pub sub topic and pops up recipient, and basically that has been done in in this issue or nipple request for that issue.

H

Are there any questions about that.

H

All right, if not, let's go on with the next one, so that was also quite a big one in terms of the of the code that needed to be written. But from a user perspective, uh not much has changed, but this issue was actually the implementation of the delivery assistant or the delivery workflow for captain zero eight.

H

So now you can use the delivery assistant and the pro the approval task within your task sequences of the of the new shipyard file, and that also means that an approval doesn't necessarily have to be a task within the complete, continuous delivery workflow. But now it can just be an isolated step within a within a task sequence.

H

So, for example, if you want to change the artifact delivery task sequence within the dev stage and add an approval, you know just have to uh add a new test, for example after the evaluation, with the name uh with the step name, approval and set the required properties. So the approval strategies for past results and one warning results, and then uh the shipyard controller and the gatekeeper service will take care of handling this task and, of course, using the cli and the bridge.

H

You will also be able to to send approval, finished events and decline or approve approvals, all right that was almost it. uh One final pro request, uh just a very minor one in in terms of the functionality that has been updated, was a request to the dynatrace service.

H

So here I have added uh support for the new cloud device. We are going to be using within captain08, so we now have uh we're going to have new event types and not uh gonna use the old event types anymore and also instead of uh cloud events uh following the spec version, zero. Two we're now going to be using cloud events 1.0, so that has been updated in the dynatrace service, and now I can handle those events all right. I think those were all for requests.

H

I wanted to show you today and I'm handing over to johannes.

F

uh Florence can you quickly jump to the working items, because you also has two items there. Two bug fixes yep that one the.

H

Yeah uh yeah after the.

F

B

H

uh If you've done it on windows, it showed you a command that you can execute within your shell in order to quickly authenticate at the captain api and, of course, that was tailored for unix based systems and not for windows, and therefore we have removed those commands and instead added a link to the documentation.

D

If I, if I may just add something to this, as we do have some people with windows trying stuff out at the moment as well, um what we've seen so far is that the git bash seems to be working nicely. So this is not 100 guaranteed that everything will work, but most of the commands seem to work nicely on the git bash.

D

I think that's something that most windows, users will have at hand.

D

Or at least it's quite easy to install in the worst.

D

H

Yeah all right and the next one um in the current version of this hero, 73 release branch, I've introduced a bug, apparently that caused create project commands to fail in case you have provided upstream credentials and that has been fixed and now I think I've done all four requests.

H

Thanks a lot, I didn't accidentally file a new one already all right. Now, I'm going to be handing over to johannes all.

A

Right, may I just uh say one word: thanks uh florian for presenting and actually uh god tight was joining the meeting, but uh it seems he dropped already, but also things got tight if you're watching the recording that you also helped us out here.

F

All right, yeah and the issues that I have to present this week um are two. Actually I don't want to talk about the first one. It is still an open um pr where I ask for review it's about the requirements that are needed to implement uh captain service that is running down in the execution plane there.

F

We provide now documentation and when someone has time, please take a look at the documentation and leave a comment in case you were missing information, but what I want to talk about today is that airman and I we are working on wireframes for the sequence screen in the bridge before jumping over to those wireframes.

F

I want to briefly open this slide just to explain why this is necessary to have a new, dedicated um sequence screen um yeah, because with the new shipbuild format, we will have um yeah, see task sequences within each uh stage and each task in a sequence like here that one has a name um that is artifact delivery has a trigger and then a list of tasks and for all of those tasks.

F

The shipyard controller will send out the trigger event and also as started as well as a finished event, uh will be create generated by the executing services, and this now creates a lot of events that are coming up in this particular case. We have five tasks. uh Multiple three events means that we have about 15 events for just executing this sequence and with more and more services that are listening to a task.

F

More and more events are showing up, and now our or not showing those events in the current uh bridge would then mean that we have to that.

F

We will fill up this right um list with more and more uh events, and that now becomes quite challenging to understand and to read- and this is the reason why we came up with a new idea of visualizing sequence sequences, and this is what I want to show you now, and what I'm presenting you now is is a wireframe where I changed the labels here in the menu through icons. Please ignore them for now.

F

I need to ask on feedback on that, because maybe they are not that intuitive, that I thought at the beginning, but the first one is for the environment screen the second one for the service screen and the third one is for the sequence screen we are currently looking at and when you click on that, you see here a filter component.

F

This filter component allows you to filter sequences based on the on the services they work executed on then also in which stage they were running, then the sequence type whether it was deployment remediation or evaluation, and then you also can filter based on the status which is active or failed or succeeded, and in the middle or in the center of this screen. You will then get an overview of all the task sequences that were executed in this particular project. In my case, in the stock shock project.

F

In this version, I'm showing the sequences here vertically and also horizontally uh starting here on the top left corner. This one is the most recent task sequence that was executed. Then this is the next one and so on and so on.

F

But I also got feedback that it also or that it might be more easier to understand when we just show them in a list where we have the newest sequence on the top, and then they are ordered according to their uh time. According to the time they were executed,.

D

So essentially the first screen is you read it like a book from left to right, then next line, um or is it better if you just read it from top to bottom, has.

F

Someone feedback on that I mean, would help a lot.

A

I think it really depends on how big these tiles are, if it. If I take a look at right now, it looks like uh it's: some space is lost on the screen and we we don't use all the the space we would have on the screen.

A

But if there is anything else that should be displayed here, then it might make sense to do it like this, because it it's.

H

A

um But I think it really much depends if we, uh if we want to to share if we want to use the whole real estate that we have or if we just want to to only.

F

Show some smaller parts of it yep, that's true. That was maybe my mistake that I did not change to the size of this tile, but it will not be that big. It will be smaller in this case to get more sequences listed here.

F

That is definitely something to be changed, but yeah.

A

A raise of hands from the community uh so uh richard, if you want to share your thoughts, please go ahead.

B

Yes, I'm you know I, I know I'm being really bold here, this being my first meeting and then weighing in on something like this, but if I may um you know, if there's more information on on this particular screen, if you were to click on one of these cards and then present more information, like some analysis there, you go yeah, okay, so that's different. This to me makes sense, um and I like how, when you click on the card it, it is noted that you are looking at that one.

B

um Is this also used for for um looking at what stage you're you're currently in or is this a configuration screen.

F

No, uh this is not the screen that is telling you the status of the stage that one you will find here on the top, but this screen is really showing you which sequences were executed in which state as stage here, for example, we have a deployment that is currently running that affects the card service and this card service is currently going to deploy it in depth and in staging actually in dev. It has finished. It started at 7, 43 and now it's finished, there's a reason why the icon is green and staging.

F

It's currently ongoing the deployment um yeah, as indicated here by this loading icon, and when we click on staging, then we see the tasks that are executed there and the task that is currently running.

B

Yeah, do this make sense to me um when you first come to this particular page, I would suggest always having one of these selected. That way, you can see that that um that there's something that is, that is being shown on the right and it doesn't look like just empty space.

F

Okay, good idea.

D

Thank you yeah. I was just going to say: I've raised my concerns about the left to right and then next line, et cetera in in a preview that I've been given by johannes, and hence we have this meeting now. So my in my opinion, this is it's good. If you only have five entries, but if you have 500 entries, I would rather just have it top down and scroll through.

F

Yep, that's true valid feedback things I mean, um I will just click on another item that one for example.

F

In this particular case, we see a deployment that is happening in multiple stages, dev staging production and then here also there's a rollout to um yeah, more clusters or more environments, in this case a production environment, a and b.

F

But still the concept is the same. When you click on one of those stages, you get then the tasks that were just executed in this particular stage and the time also maps. Here we see that it started at quarter past eight, and this is then, where the approval came in at quarterback state, then the deployment succeeded, the test succeeded, the sli were fetched, the evaluation was done and finally, the release happened in this stage.

F

Just one final click: when you have just an execution of a sequence in one stage, then yeah you will see just the stage that was um affected by the sequence and yeah again the tasks that were executed.

F

All right, yeah, thanks for the feedback already in case, you want to provide more thoughts on the screens. Please just jump to them to the issue which I posted here in the agenda and just leave a comment there.

F

You get an access to the wireframes by clicking on this in envision.io link and you can click through the screens, and I also opened this question whether we should go the way that we present the tiles in a book style where you read it line by line or in a in a list style where you have it stacked here in a list format.

F

Please just um also please leave a comment here, all right. That was it from my side and I'm with the last one in this list. Then it's time to jump over to the working items just very quickly. Let's go through that one that are ready for review, but have not been presented that one came from ankit.

F

He did an improvement of the install command of the captain install command because um yeah you can brought up this idea of showing already so when we are installing captain with a load balancer or with a node port. Then we know the the end point uh because yeah it's in the cluster and the idea from viewing was that we then can provide this information right here with the captain.

F

Aus command, then with the value of the endpoint, as well as with some command to get the api token, and this was implemented by ankit in his pull request. Think this was not the wrong one.

F

Jump back, I want to go there to to show you one thing.

F

Yeah I implemented it, but we discussed that we also need to add a new feature flag, a new flag which is called the height sensitive information or height sensitive data. Because with this new feature, the captain cli command is showing you sensitive information. The api token, as well, just just the end points. The api token will not be shown, but the end point is displayed and in case you're integrating the captain cli in an automation tool.

F

Then please be aware that you can use this flag to hide the api token, and then it will not show up in the output yeah in just to stay safe and to to don't leak, sensitive information about your cluster also, but the feature itself works pretty cool and it's it's a quite handy enhancement.

F

Then I mean those two pull requests are still open and christian. What is the status of my creating to travis.travisci.com.

D

As far as I've seen, it's done, I'm just leaving it in ready for review in case we're getting any news or any updates uh by our end users, but it should be done for now, so I've migrated captain captain, contrib and captain sandbox. um I think I need to migrate. Captain example still I'll take a look.

A

D

Verify that for captain.

A

Sandbox, it's working because we just deployed our or released the amsoil tower service today, and we did not change anything and it was working right away.

D

Well, thanks to travis that the migration is working.

F

Cool then, in progress we will not go through this list. There's just one item that I want to open. It's that one, because it's required for 0.73 is tight on the call today.

A

I think he rejoined. uh Let me find him in the list and enable his microphone, and so I tied a few joints. Then you should be able to unmute yourself and um yeah join.

F

I

How are you doing yes, okay, so uh I want to talk about the first eu, the timeout for this one, okay, so.

F

I

You you said that I should um try and check the uh try and check the the code a place where it displays the additional information, but I think there's a recorder ready from line 161. Maybe can I share my screen? Absolutely? No. You can okay, okay, I don't waste so much time in this.

I

I always start with that.

A

Sorry, I I will promote you to panelists, so you can share your screen. uh You will rejoin in a second okay.

A

Okay, uh you should be able to share his screen just in a second uh yeah he's back and he should be able to.

I

Share yes right here, so uh the the issue is that the person was not able to uh during download the the download button times out.

I

So this was um the uh the code I added to it when you said that um I didn't have to use the time, and you said I should check out this um upgrade chat code. So this is here right here. So what they're? In the code there's already a template for uh an error message when you and when you don't have to when you're upgrading everyone installing a grid like what is that a problem there?

I

So I to the best of my knowledge, I think uh the problem is that there needs to be some kind of time out during that hanging session. I don't know the best way to solve this, but looking at what you said, uh this is the things I've found out during the the process of you know. So I think there's already a solution in the code.

B

Except you want.

I

To me to have extra information that which, including.

I

Including adding extra space for the port yeah, so that's what I found from here. You can go ahead and you want to make a.

F

Comment um tight: can you jump.

I

F

To the issue, please, just just okay, the issue.

I

Okay, this is the issue right here uh turn out.

I

Wait. Sorry, sorry, I'm taking some more time.

I

I

Stairs, I'm really not sure how to look for this right now, um turn out.

F

No, let me let me just give you okay, let me check.

I

I think these are newer issues.

A

If we can find the issue number, you can just click on on any issue and uh then in the url we just change the issue number yeah. This is oh, you found it perfect.

I

Yes, so this is the issue right here.

I

Okay, as in enough resources and when you try to install captain at some point, the insulation is stuck and nothing happens. So the the idea is that it wants that. I started not knowing. That is enough to my fault, because.

F

At the question is where we can get more information information. Yes, on the code that we can uh present to the user. As christian mentioned here, um that the upgrade chart function is performing the upgrade and when this one fails, we should provide an indication why it failed, and this is actually the goal or the the.

I

Task yeah, so he said that we should provide more information on why the upgrade field, or we should the the upgrade to time-outs.

I

He said looking at this very well, oh my gosh. Can I also consider, mr for this.

I

I

So yeah, so you, you just said that should um okay, this is next, so we don't waste too much time. That's that's the that's the issue right there, so I I tried to implement a simple timeout. Well, you said the time was not.

F

I mean the timeout um it's kind of risky because we setting a timeout, it can work, because now you set a timeout at seven seconds, and this is from my point of view too low, but setting it to 20 or 30. I don't know whether this will work.

F

Going with a timeout is, I think, not the best approach, it's more about providing more information to the user, why this upgrade failed and did not work the way it should. Okay,.

I

Yeah, that makes sense, okay, so uh about um about the uh the providing more information to the user. Okay, why am I here? uh I I should I dropped some like yeah, so I was saying that the the upgrade chart already provides. So I'm thinking that should I just have some more some comments here that you could feel doing due to some.

F

I think I think the best approach is that you um that you try out what happens when the upgrade fails, because you have a too small cluster just spin up a cluster with with one node and a with two, um with with a just with one cpu and and not that much memory, then you try to install captain there and what message do you get when the upgrade fails or the installation fails?

F

And this information should then be presented to the user.

I

I

Yeah, so I will try try that out. Yeah.

F

Okay, all right then ty. I have another question another question: um do you find time to work on that issue? um The the the other one right.

I

Yes, uh yeah, I I had some questions on the um like. I, I was trying to um look for a way because I don't have captain installed. I've been having some issues installing it you know so. I was looking for. I asked that whether I could just install the bridge service and the lighthouse, of course I've been trying trying to uh to do it on my laptop. So um I think juban sent me some um information on trying to install it on a box.

I

B

Working on that.

I

Too, but I I still would have preferred the way I could install the street on my laptop like on my local host, so I can test for changes. I really don't know how to do that yet get more resources.

F

Okay, um ty there's just one one thing that um I don't want to take it away from you, but we want to go. We want to release 0.73 next week and this feature should be in there and in case you don't find time to work on it. It's totally fine and then we will just uh take it over.

I

Yeah yeah, it's okay! You can just online.

G

Maybe just a side note for everybody who is going to um work on the bridge in the future. In the repository in the bridge directory, we have a readme how to start the bridge uh on localhost, so you just have to provide the uh remote url so uh url to the api and the api token.

G

So you need to have an captain cluster somewhere deployed and then it's just uh um node or actually it's an npm uh script that you start and it starts the bridge locally and you can develop on it very easily.

G

Just as a side note here,.

B

F

Okay, uh tight uh when we will start working on that, we will let you know by a comment on this issue: okay, just okay, all right, perfect, then those I will skip and we'll jump over to the backlog in the backlog. We have one issue for the statistics service. The statistics service is running into memory leak according to a monitoring solution, and here we can see that it's constantly increasing the memory and there is no drop um yeah, and this fills fills up the the memory consumption of this service.

F

Can someone take a look at this problem.

H

F

H

F

All right thanks a lot. I think I'm not administrator of this repository. um Maybe flooring. Can you assign yourself to it by.

H

D

All right, I was going to say I can assign it later but yeah sure, okay,.

F

And the second one is that the integration page and the example is not working uh because of a wrong time format, because when you go to the bridge to the integrations page there, you find an example for an uh evaluation use case for the quality. You quality gate use case. But when you execute it, it's not working because the time frame is not udc, it's another one, and then you run into a problem that you are actually trying to evaluate a quality gate that is in the future.

F

I also consider it as 0.73 related um meaning that there is a little bit of pressure on that one. Who is up for working on that.

G

Yeah, you can assign it to me since it's in the bridge, okay,.

F

F

Oh okay, then the next one that.

G

Came out started already working on that one.

F

G

It's uh critical and blocking for.

G

One customer and yeah, therefore yeah you can assign it to me.

F

Thanks a lot yeah yeah. The problem right now is that the bridge is showing just 20 services and in case you have more, uh they will not show up because of a pagination um situation in the configuration service, then that one is not that critical but pulled it in just in case um we can well. We have time to to work on it before we starting the 0.73 release.

F

It's about a problem that that appears when you have an slo file where the um the value of what's the problem, but what a value of include result with score has no quotes but is shown like here, then the bridge cannot read this information and is displaying it. It's actually not showing this information, and the fix would be to to how to read both versions of a json file of a yaml file, one that has quotes and the other one does.

F

F

Yeah you can assign that one also to me. Okay, thanks.

F

F

Okay, perfect and then I have yeah the the issue about the 0.73 release, we're going to start next week. I did not fill in or did not add the information or did the checklist there. It's just a an empty box right now, but um yeah by next week. We are working on that one and our par apart from the release itself.

F

I just want to point out that there are still a couple of issues in there that are regarding the services for of the contribution of the captain, contrib repo and the captain sandbox repo, because we need to up them those in order to make them working with the new shipyard format and also with the new shipyard controller.

F

Actually, this one summarizes all the services that need to be upgraded.

F

If you have time, please feel free to pick one of those uh issues and to help us upgrading all the contrib and sandbox services to make them older, eight compliant- and here are all the issues um further down.

A

Is there some kind of test framework that we know that they are 0.8 compliant or how can we validate um that? We made the changes correctly in sandboxing contrib services, because.

B

We might not be.

A

Familiar with all of this, all of these services.

F

That's actually a good question. I mean no. As of today. We don't have a test framework for that, but.

F

We should consider of providing one just to make sure that the implementation is is working uh with auto data yeah. So.

A

Something like uh the test framework should just send all the different kinds of events that are possible within and send it.

F

To the to the service.

A

And it will check if it receives the correct server.

F

That's true: that's.

A

F

Everything kind.

A

F

Very simple shipyard controller that sends out all the events and is listening or waiting for a started and finished event. That's basically, it.

A

Yeah because I I think in the beginning most of these services, they would not either would not uh understand the event that they are receiving or would fail in sending back to the correct event. And so we can check easily if everything's done correctly.

F

Yep, that's true: um let's have a follow-up discussion on on building such a framework.

F

Okay yeah, those are then maybe more for auto8, which we will um discuss next week, all right. That was it from my end,.

A

I just realized that we already took a little bit yeah.

F

A little bit over time.

A

A

Do we have any more uh items on the agenda for today, but I think we are. We are good, then one last item from my side, so we are also working on improving the captain website.

A

If you have any feedback that you want to share, please let us know- or let me know in the caption slack um and uh we are happy to to discuss how the the new iteration of the captain website should look like um then uh yeah thanks everyone for joining sorry that we took a little bit more of your time but yeah. We will see each other next week and uh please note there are two captain community webinars also next week.

A

If you want to join them, uh the links are in the meeting, notes and we'll also tweet about this next week and have a nice evening. Everyone and see you all next week, bye thanks.

B

A