keptn Community, 27 May 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keptn Community Meeting - May 27th, 2020

Description

May 27, 2020
Keptn Webinar later TODAY: “Autonomous cloud journey at amasol with Gitlab and AKS” https://dynatrace.zoom.us/webinar/register/WN_bNg36mO1SzCGwETHNB5uTw
Noteworthy Issues & PRs (Review)
Florian:
Fixed configure bridge command
https://github.com/keptn/keptn/pull/1839
Upgrader for 0.6.2 to 0.7.0
https://github.com/keptn/keptn/pull/1826
Ermin:
Filter events in list of root events Filter events in list of root events · Issue #1342 · keptn/keptn
Christian K.:
Add testing for keptn uninstall/cleanup of GKe resources https://github.com/keptn/keptn/pull/1837
Fixed integration test for quality gates
https://github.com/keptn/keptn/pull/1842
Christian H.:
Thomas & Andreas.:
Issue https://github.com/keptn/keptn/issues/1822
Research for splitting control- and execution plane https://drive.google.com/file/d/1I_Oem7f-XIEB9WImZ9U3Rag10ivgM8AB/view?usp=sharing
Johannes:
Open PRs on: keptn/keptn.github.io
Roadmap → KEP / SPEC / Issues:
KEP 09 - Spec change of remediation action config
SPEC 31 - Implementation of spec change as proposed in KEP 09
TBC: Keptn Core (MongoDB/configuration-service): Manage open remediation workflows in the materialized view managed by the configuration-service
TBC: Keptn Core (remediation-service): React on problem.open and process predefined workflow: trigger action, wait, evaluate, continue remediation or send a remediation.done
TBC: Testing: Platform/integration test for self-healing use case
TBC: Testing: Integration test for use-case of scaling a pod based on a Prometheus alert

SPEC 30 - Add ProblemURL as optional field to problem.open event
#1694 Add ProblemURL to the keptn problem events

Working in progress:
https://github.com/keptn/keptn/projects/29
Backlog Refinement / Next Sprint:
see 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
Attention: new dates for upcoming community meetings
Thursday, June 4
Wednesday, June 10
Please subscribe to the community calendar to stay up to date (ical)

Learn more: https://keptn.sh
Get started with tutorials: https://tutorials.keptn.sh

A

Okay, welcome everyone to this edition of the captain community meeting at the Wednesday its yeah. Today we are meeting on a Wednesday, because uh huge parts of the captain team are not available available tomorrow, so for today, I just want to start with the short announcement.

B

A

Today we have the captain webinar on the autonomous cloud churning of one of the early captain adopter and they are also using it lab and a guess. So we will have a captain webinar. You will find the link to the web thing having our webinar in our captain community meeting notes. So in this document you will find a link here. If you don't have access to this document, go to captain dot, SH go to the community and you will find the link to this document on our community page or just join us in the captain.

A

Slack Channel and you will also get all the latest information in the captain slack. So you will stay up to date. So I think that was basically the first and only announcement and with this I'm already handing over to Gloria to kick it off with yeah some some code of captain. So please Florian go ahead. Alright,.

C

Thank you sure my screen.

C

Can you see my screen already? Yes, okay, all right today, I want to show you a to PRS or issues. So, let's just start with the first one, so the first was actually a part that noticed which is related to the new cap exposed bridge to month. So basically, what happened here is very good summary about the problem here. So basically, what happened was when you used the captain configure bridge and to expose the bridge make it.

C

When you had the full installation of captain using the Easter virtual services installed and in parallel, have existing nginx ingress installed, which was in the namespace ingress nginx, then this command left it with invalid configuration because it wrongly assumed that captain also uses the nginx ingress to make the services available by the internet. So the fix was to check if the ingress nginx need physics is, and additionally, if there actually is a captain ingress within that namespace available. Because that's at a point when we can tell okay, captain actually uses nginx.

C

So the code for this fix is in this pull request, and here you can say, see exactly what I've just described. First, we're getting the Cuban Eddy's library plank.

C

Sdk and we iterate over the namespaces within our cluster and here, if we detect an ingress.

A

Index namespace.

C

Read in addition now check if the ingress for the captain installation are actually it's available.

B

C

A

Like to to add a discussion to this, actually the implementation is fine. So that's that's good! That works I, think, but going forward. We should probably not have the you shouldn't have to separate the how we expose captain's bridge and captain API, where it's or nginx, so the nicest thing that we could have in the end would be to reuse Q&E. This ingress at some point, so not to have this distinguishment between east you or nginx, every half right now. Maybe that is something we can have an issue for in the future.

D

Yeah I already did some research and I used an ingress object in combination is also a community's ingress object in combination with SEO and I ran with our cube. That missed our estimation in strange issues. So here we definitely should make new research issue when we have installed su-15.

A

Absolutely because I also know I think Tomas can it I can add some info internet, but Tomas has tried it out with traffic I think and the English objects that you already have seems to work with traffic or traffic I. Don't know how it's pronounced, so it.

D

A

With traffic is the.

D

Only difference is you're, always in the metadata, and how much can you please move a.

A

D

So this is true. This would be the best case to only have one ingress, which can then be used with nginx. Is your traffic or any other provider? Yeah.

A

But yes is that we will have to do some research, so okay, that that was all I. Just wanted to add this to the discussion now that we're in this big round of so many people here, I.

E

Have an additional note? Actually, it was Patrick who brought up this problem, who or and who identified this problem, and he is also on the call: that's pretty cool Florin. Do we have an image or is it it's merged right.

E

Okay, okay! Well then, it's also possible that we can send Patrick image so that he can dry out this bug fix yeah cool.

E

A

Who was the fixed in the API or was it fixed in a CLI.

C

A

It's breaking I, think I'm, not sure. If we can just, we can try uh-huh uh-huh. That's.

E

True I forgot about that and it might break it might.

A

Break helm version three and stuff.

E

Okay, all right then I Patrick I will send you an image and then it would be cool to when you can try it out.

C

Which is again upgraded which will be used for upgrading captain 0.62 to two zero seven zero, because in the upcoming version, we're going to be changing a lot of things in terms of how we structure our data within the.

B

C

Baby, so basically, what we do is, or what we're going to do in version 7 we're going to have a separate collection for each project, which will contain all the events for a particular project and we're going to have this materialized view that is used by the configuration service and that we are already shown you in the last few community meetings.

C

And that means that when upgrading through this new version of captain, we will have to do some database migration and that's actually what this is upgrade will do and also it will do the migration to helm version free.

C

So let's have a quick look at the code for that I'm. Sorry,.

C

Again, the way it will be useful be the same as.

B

In the last piece of.

C

Your operators, which will be humanity's job and records.

C

Yeah, so here it is so. First of all, it will of course apply.

A

C

The relevant manifests and upgrade the images to the latest version, so that's not too exciting, but actually the magic that mean operator is here in the first part, so it will basically collect all the pre-existing events of the manga repeat: data store, extract them and based on the project there are part of will insert those into the new dedicated collection for that project, and also it will go through all the events by event, sourcing mechanism. It will construct the materialized view for that project and the services within that project.

C

So there will be no manual migration of your pre-existing events necessary, and you will have that overview of your past captain, we're close after you to the newest version.

E

Not about that, but you are also doing the helm upgrade here what's relevant for for this step, I.

D

Did this, and and what we are doing here is we are upgrading and to releases, so helm works in releases into him releases, and this is the magic which converts old. They hadn't releases, which are yeah, which are managed by Kevin. Okay,.

E

Cool and RJ changes required in the help chart no.

D

E

Okay, this means that we can actually leave the helm like they are, but we provide this feature in the upgrader and this will then arm. Then we are help sorry compliant I.

D

Mean we, we can also update our examples, but we do not have a backward compatibility. So therefore, I would work with.

A

C

A

Locally right so.

D

As long as the user doesn't touch him, he never needs him actually and but if he would like to list all the releases.

D

A

We should make sure in the release, notes that there is a big bullet point that state a we have credit to help version three, if you're using help version two or wireless Eli or whatever, then you need to upgrade, but as far as I can see, I'm not sure like are you distinguishing here in the script between helm version two and helm version 3? Why are the name or is that automatically because helm version 3 has a different name for a command-line tool right.

D

In the image I've installed, M 2 and M 3 and yeah for creping the releases I'm still using them to version and then I can use M three in order to upgrade this to converts this. But during a talk, IDs I actually identified a potential. We pack, because here we are in line. We are wrapping all three pieces and I'm not sure, but we should really convert all releases.

D

Actually, we should only convert those releases in which can be managed. Oh.

A

Yeah and I also see something in line number seven: the hell minute command I know that this will fail. If the cluster is air gapped and does not have a working internet connection, in fact, I guess the upgrader will fail anyway, because we cannot install the newer version of helm. If we are air gap, can we nope.

D

D

A

D

Line 9 should be corrected. How can we do this?

D

So actually we would need the name of the project and the shipyard right.

E

Or is the name of the project sufficient.

D

Let's discuss this in.

E

An issue and because.

D

We have to do some research here.

A

And I: ask you another question unrelated to the upgrader: does it only work from captain 6.2 or also from captain 6.0 or 6.1? This.

C

One works from zero: six, two, two zero seven. So in order to operate from zero, six zero, two zero seven, you have first apply the operator job for zero six one and in the operator, top four zero, six two and then so. Maybe we can also provide a combined given at this job that basically changed. Those.

A

I think it's fine if the use I have to apply them manually, because there's so many things that can go wrong. It's better to apply them step by step.

C

All right, let's move on I guess, yes,.

C

F

You very much Brian I.

F

Plan to present only one issue today its filter events and lists of Route events and a bridge, so the issue or the feature request, was to allow filtering of events by selecting certain event types shown in this wireframe here.

F

The idea is that you can filter by those even types so that you can easily, for example, detect or list only problem detected. Events in if this list is getting longer so I'm going to do a live presentation on my local machine. Actually, since I had some issues with some npm modules, which broke all my pull requests that are currently open. So when you click or when you expand the service, you get the full list of all events since currently, oh and I just realized that there might be a usability issue since yeah all event.

F

Types are pre-selected by default, but you can disable certain event types. This list will grow once you have more event types here in this list on my local machine. I have just this configuration change and service create events at the moment.

A

I'd like to ask a question that I, probably.

B

Already and so.

A

On when new events come in, while you have this open, does the list also update with the filters yeah? So if now, yeah, okay, good.

F

Yeah, since since this stream is yeah- and how is it called- is a is a subscribe, I'm subscribed to to the API and with the response every time yeah, this list is populated, and this one is also updated on yeah since I.

B

F

To check yeah, actually, the filter is working, client-side, yeah right, so all right, cool yeah. The the only thing that is probably missing here and could be enhanced in the future is that this setting is not saved. So we don't have. We don't store the.

F

Check box State in the database, nor we do on the client side, since the option would be only to store it as a cookie or or in the session storage, and for that we would have to introduce cookie policy, which was not at the moment planned. But in the future. We have another feature on the list which will include a cookie policy and there we could yeah yeah I, think that feature I think.

A

It's okay, I know, Enterprise great services that also don't store, filter options because of reasons yeah.

F

E

Yeah further questions, yet just a minor can we remove the border around on.

F

The button from.

E

The bottom yeah I.

F

Will have to check that one, because this is a button from the barista components, library and yeah. I have to check if there is an option, because there is actually a configuration possible- and this is the lite version. Otherwise it would have the background in the color and I cannot be white, but I have to check if yeah or I cannot.

A

Yeah, maybe it's also maybe it's also possible to not use a button, but just a hyperlink and anchor tag or whatever I don't know what barista actually offers so I.

F

Will check on that yeah mmmm.

A

F

Yeah and yeah, all the other improvements in the bridge actually are just bug fixes which I'm not so that's from my site. Handing over to.

A

Thanks so I'm still actually working on issue, 1 7, 7 7, basically making sure that captain works on mariah's kubernetes versions and I'm. Implementing integration tests with and I always find some smaller issues to fix and I'm presenting some of these, so the first one I want to present is one, that's actually very interesting, I think for the community, so we're running the platform tests on Google communities. Engine I noticed that there is some orphan resources left behind like persistent volumes and load.

A

Balancers IP addresses firewall rules and stuff, and the reason for this is very simple, which is called Chi cloud. Cluster delete or chi called cast. The container delete, don't know what the actual command is, and this does not delete persistent volumes and load balancers.

A

But when we run captain install and when we on board soft shop or a cots micro service, we actually do create bases and volumes and load balances, so very easy fix to remove some of these orphan resources, or at least to remove them to avoid having them created, is to run captain uninstall.

A

When we do our integration tests and two more fixes would be to remove all services of type load. Balancer I did not do that for the reason that our integration tests also work with a pre-installed version of SEO. Basically, the version that Google is installing, so if I would remove the load balancer of the pre-installed version of SEO Google, which is reinstall it and I'm not sure what kind of havoc test this would be causing. So I did not do that, and also removing all persistent volumes.

A

I also did not do that, but what I did I will show you in a pull request. Basically, I have a test called edit the test or integration test. That's called test, captain uninstall, that's basically also. We have an integration test for Captain uninstall, where I verify that everything from a namespace captain is removed. If the namespace still exists, something obviously hasn't worked and I'm exiting with one.

A

So the integration test will fail, and once this has been done, I'm also making sure that I'm deleting any namespaces that I've created during the previous steps in integration tests, so basically I always create a project with three stages: Def stay staging and production. Therefore I'm making sure that these namespaces are always deleted, and in case these names will just don't exist for whatever reason, I still want this command to succeed.

A

So I added a pipe true behind that and yeah with that little script here, I'm 99% sure that we will not have orphaned resources, or at least we won't have many. But I have to monitor that. Obviously, but yeah lesson learnt is for everyone, that's using captain in an automated way to always clean up afterwards and the second thing I've also added with this pull request.

A

If it used the disk size required when creating a cluster from 100 gigabyte to 25 gigabyte, because this is only the disk size of the boot disk of the kubernetes cluster, it's not the disk size of the persistent volumes. So that should also like save some resources and I'm trying to use preemptable notes now, thanks to Thomas for this tip by the way, so preemptable notes cost like 25 percent of a normal mode, and they can be canceled any time by Google. So if Google says hey, we don't have a note for you.

A

We'll take the note away from you. That would be bad for us, but for an integration test. I think that's fine, because in the worst case we will restart the integration test and yeah. It just saves some resources save some money, and this probably also good to know for other people. If you ever want to implement some automated tests with captain or g-cloud, you can use that kind of setting.

A

Right any question regarding this pull request.

D

Not regarding the pull or twist, but if I delete trusting using the chill out on website. Yes, then felons committed or not- and also the PVCs, of course so.

A

From the clean up that I did in the last couple of days as far as I can tell no, because that uses the underlying API, which is the same that Ritchie Claude seal I, would use so there's no difference as far as I can tell between the UI and CLI okay.

D

Good to know, actually, this is exactly the same for EWS. If you create cluster eks cluster, using the ETL CTL and then add manually at node balances, they are not cleaned up by their underlying Kara from script and.

A

Yep I'm pretty sure we can have a long talk about resources, not being cleaned up. I already talked to like so many people that verify to me: oh yeah, that's a that's a huge pain, so yeah. The second thing I noticed, is that our integration test for quality gates, sometimes failing I'll, actually show the result of the test. So actually I'll show the test itself I've Travis over here.

A

So one of the evaluation done events that we verified in integration tests were given very training violated. True for response time, P 95 and the reason for this is. We were checking that the target value of response time P 95, hasn't changed from more than 10%. So it's a relative change and the value was always quite low. Around 30, milliseconds or 32 milliseconds and obviously 10% or 32. Milliseconds is only 3 milliseconds and if this goes at 36 milliseconds yeah, we violate it and that just happens randomly. This could be a networking hiccup.

A

This could actually be on purpose. We don't know an easy fix, although was to increase the the target value here to 20% and I'm, actually showing you the failed test here, which is marked as red icon in Travis and with the green I. Can here's a positive test so here the criteria I've increased it to 20% and it's no longer violated, we've actually, the same target or the same value or almost the same value that we measured in here with 36 milliseconds.

A

We had 36.3 or something also in the other test, and the changes were also minimal. I only had to edit our our test, SLO files and I've added the 20% criteria instead of the 10% criteria here, and what I also did in addition, I edit up warning criteria so like to make the test a little bit more realistic. We could say anything between 20% is fine.

A

It's an automated pass and 50% would still be okay, but we should receive a warning and if the response time changes more than 50%, then this could be considered to bad user experience or whatever. So we want to be notified so yeah. The idea is that you can reuse the SLO files that I have in my test assets folder here for obviously installing it on your own or, if you just want to yeah, also play a scenario like this. With your tests, you can we use this as a loan files and yeah.

A

Alright, any questions regarding this pull request.

A

Okay, nice um yeah then I'll hand over our, not sure. If is Christian Hoffman here does he want to present something he's not on a call? No all right, then I'll hand over to Thomas and and yes.

D

Thank you so Thomas and myself. We have worked on a research issue. Well, we would like to run the captain execution plane on a different cluster compared or compared to the other class running the Capricorn. So technically, what we have done is we have put the chavita service in the new cluster and then sure thank you. Technically.

D

We have put on the Tomita service as our research example in a different server in a different cluster, and then we introduced a little side cub, which holds whether there is an open test which should be executed by the imprimatur service and this almost the outcome of our research with it. So there is now a little prototype available which allows to run to meter not inside the cabin control train in that second core, that from my side- and next is johannes.

E

E

Actually, I don't have any pull requests to talk about, I have more or less pull some work to distribute when it comes to pull requests because can I ask someone of you to take a look on the open pool request for documentation. There are four PRS open and there is still approve all missing.

E

The first one is for the upgrader, actually that one florian showed us that one when we explained how to upgrade from ODOT 6 2.7, then here it's a pull request from christian about simplifying the use of mini cube, and then I added a list of taka files that we actually installed when we do the manual install so that it's obvious, which docker images get applied. And last but not least, this is an this- is the community support matrix that shows the support of captain or the support of communities with captain or at 7 I'm.

E

Pretty sure that we need to update update this one as close, we get to the release date, but nevertheless, I want to show it on the develop folder, so that folks out there can already take a look at the support metrics that it's going to be supported in the future. I.

A

Think that also goes to my ticket a little bit so I'll notify you if there is any changes, but so far it's looking good for anything between 1.14 and 1.18, and I hope that we can release captain before Q and it is 1.19 yourself.

A

E

Must must go fast now, no just kidding who can give me a review on those pull requests.

A

I can take a look at it tomorrow. All.

E

Right thanks a lot cool I mean well.

A

Well, except for the one that I opened, always correct, yeah I,.

D

Can take this okay.

E

Thanks a lot then I will go on with with the next point in the agenda. It's about yeah. What's the current focus of the of the development train so to say Oh what what is the Fokker --nt, what we are currently focusing on? It's all about the spec change and the captain enhancement proposal for supporting custom remediation providers.

E

Yeah last time, I showed you, the company kept enhancement proposal and I've broken it down into a spec change, and now we have also broken down the issues here. I have marked four of them as purple to be see, means to be created, I mean we have already specified those and we are ready to create concrete github issues.

E

All right yeah: this is a focus we are currently working on, and this brings me now directly to the booking items for the last week and let's switch over to the Kanban board and like we did the last few community meetings and deaf meetings. I would now go from the right side to the left and we yeah everybody I just says short sentence or short yep statement about the issue and then a weekend all we know the status of the issue and we can move on to the first one program.

E

You already showed us the Patrick's.

C

E

Cool then the upgrade mechanism we also discussed lengthly and then show usually as a live demo or the code, and then we have to pull requests or issues regarding the bridge, I mean. What's there.

F

Actually, there are two.

E

F

Are actually yeah done, but I have broken built there because of some NPM issues. Actually, I have an issue there with the bubble transpiler for typescript there there is an issue and I have to I, think I have to upgrade or update, and the diversion there and yeah I haven't managed to finish that one. So, actually all the pull requests have the same problem in the built when they are compiled with angular in angular, with the production setting.

F

So with ahead of time, compilation which I haven't used locally during development and dead raised and during the pull request, everyone.

A

That's implemented stuff with angular and has used a ot can sing you probably a song about that and has has seen such a problem. It's yeah. We should get it fixed. Maybe we can make a separate pool request based on a master branch in the movie base. We gotta.

B

A

Or stuff like that, because I think we cannot go forward without that fix yeah.

E

Can we put it on on very high priority because it.

A

E

Be really cool to to get those two pull requests into master so that we can also demo and show all your changes you did in the in the bridge. Yeah.

F

Of course, I just realized today in the morning the issue requests and it will be fixed during tomorrow. Okay,.

E

Okay, cool perfect to get those two pull requests into master yep.

A

I just created an issue and they'll edit actually into the backlog, and if it's a, if it's fine before everyone, if you I'll directly assign it to em in.

F

Cool thanks and I.

A

Can help if you need some help of I've triage, that's problems more than once. Let's say it like this: okay.

E

All right, let's jump over to the next column, it's the name, progress, column, Florian. What's the status of the unleash service.

E

And we skip this. One Kristin is not here, Andreas and and Christian. uh What's the status of this one, removing the host header so.

A

I think I have an open pool requesting go utils, I'm, not sure if I've marketed is ready for review yet, but as discussed last week, we will need that change. So I'll I'll just give that pull request open for review, and then we can yeah finish this issue, I think Tomas I'm, not sure did you also have a modification of the CLI for the WebSocket communication.

A

B

You muted, I think I only use QR your fix. Oh okay, then.

A

I'll just get that one merged, and then we can move that to ready for review and yeah should be done next week. Then I think.

D

There's also an issue in the Installer, so in the ingress objective we also have to fix it. Yeah.

A

Well, we don't have to, but we should it's yeah. If we want to step backwards compatible, we should leave it in the in the manifest, but yeah we can. We can remove it from the manifest I'll. Add that, as a note in a issue.

E

All right, then, our Christian: what's your status on the kubernetes support, so.

A

Within Travis I think I've done everything that I can do and I still want to implement a Jenkins pipeline where I've requested some support internally. That will help me to create the Jenkins problem, but I have tested it two weeks ago when it worked two weeks ago. So it really just depends if we can live with, we've tested it once or if you want to have an automated test.

A

Okay, that that is for a full installation. The smaller installation we've used, kiss quality gates is tested automatically automatically.

E

All right, and then we have one two, three, four, four four issues regarding the.

F

E

F

Issue with those I haven't opened the pearl request. Yet because of the issue with the äôt build okay.

E

F

So I'm going to fix that one first and rebase all those branches and then open the poll requests. Okay,.

E

Then the fixes some the highest priority now but short statement regarding the issues itself fix of.

F

Yeah we have two bucks there mmm-hmm. The SL ice should show success through false also for info SL ice. We had an issue there that if the SL I was just was configured just as an info and there was an error message, it was showing just to result with zero, and you know clue that there was actually an error, and the next one was breached shows wrong duration for revelation done events. There was a problem with the calculation of the duration after event itself, and actually it was just wrong yeah, creating a wrong time.

F

Difference actually I, don't know using the moment, Jas library, for calculating the difference between two dates. Currently.

E

You could solve the problem, yeah yeah. That's then the.

F

Next one was presented, filter events in list of route events and the last one unit tests for bridge is one that I yeah still had not the time to look at, since there are also some unit tests that I'm did I want to finish in the bridge.

F

Then I'll close that.

E

Alright, this brings us now to the issue. Status are signed and committed.

B

Thomas, a short short update on the first one: yes, I didn't work on this this week, but I think I will deal with it next week.

E

And this one we can you unmute, Thanks, sorry and yeah. We already discussed that one. It's the äôt build the a problem that is currently.

D

E

Blocker for getting all the changes from the bridge into master mm-hmm nice, then we can go over to the back lock and it's again a bridge topic on the top. It's about the out of sync labels in the environment, screen I, think I explained it last time and then showed it in the in the meeting. Is there some information needed.

F

Yeah I think that should be clear and since we have the environment screen already implemented,.

E

And the information will or is available in the payload or in the API in.

F

E

Okay, then, there should be kind of straightforward.

E

I mean: are you going to take this one yeah.

F

I would take this one in yeah for the next week.

E

Okay, what's it not working, not myself.

F

Yeah but feel free to join the fun actually.

E

F

E

You some code. Yes,.

F

Since you're doing really great mock-ups design, Thanks I.

E

Could also take responsibility for implementing those right yeah as.

A

Long as you don't have to deal with äôt builds trust me.

A

So yeah I think the next one I'll talk about it. Well, you honey is a science. I mean I, think that's still the issue about creating a contribution to beliefs, the components for highlighting a column in a heatmap.

A

Do we still have a timeline on that, or do we actually want to remove it from the backlog, because I think that's the third time now that we are discussing this issue and not doing it.

F

Yeah, probably we should, or we could remove it from the backlog. Since we have currently the focus on the manual approval cap and we have a sister.

D

F

Yeah assisted delivery.

D

Is orient yeah.

F

So yeah, since we have to focus on those issues.

E

We still need it for old at seven, but let's put it down in the priority list. Okay, let's move on to the next issue. It's about platform and and integration tests for their.

B

E

Case well, yeah I have written down or test scenario. That was actually the combinations that are supported with the test plan, actually with a shipyard file and containing those combinations of approval strategy and then the test scenario which they chose. All the events that should captain and then the status should be checked.

A

It's the seal I already implemented for that, because that one is the.

E

A

Item in the column, so.

E

No, it does not, because if the functionality to implement the integration test is there, the third item is just a enhancement and a more convenient way to.

A

E

An open approval.

A

I see in that case, I can do it. The integration test, because yeah all right, I've written the last one for quality gates, so I'll have a I'll. Have that one done as well.

A

Okay, then I guess it's easier for me if I talk about the next issue, so this is a back installing captain with gateway node port as an option works fine. But when we configure the main afterwards, we will break the configuration I. Think if you can open the issue quickly, there is some.

B

A

In it, I can actually request control, but yeah there's some information in it. We've already discussed it a little bit internally, but essentially there are some ways to fix this and most likely we'll have to fix it with the configure domain and within the config map and yeah. It should just include the port in the in a config map and configure domain should allow also configuring a port or automatically determined that the port is needed, or something like this.

A

Right so who wants through that.

A

Make sure to try out mini cube and not mini shift, maybe that will help that your MacBook doesn't explode or lift off. Apparently.

E

Okay, then, let's open this issue, it's a new when we derived it from the last on deaf meeting, because what I'm proposing is that we used the captain's and event approval, finished command for a specific project stage and service, and when the user does not provide the ID for the approval to confirm, then he gets the open approvals presented, piracy a lie and finally, yet a user can select an option which approval he wanna confirm and this time yeah helps or attaced and continues the delivery. Workflow.

E

Everything clear so far or.

A

Freeze issue use did a strikethrough for the option so.

E

A

That counts for a whole yeah.

E

It's it's option: a I, actually yeah I made a poll where asked the community which option we are going to implement and yeah. It turned out that it's the best way to use the Katniss and the event approval finished command, because this is the command that the user is going to use. Orey's is yeah, make some makes more sense and yeah.

B

E

All the things that are striked on it's a CLI issue, there's don't you dependency to Trek or you have the information? Is there the command? Is there it's just an a feature add-on to make it more convenient for a user to yeah, to use this use case or to run this use case.

C

E

A

Sorry, honest for interrupting, since we are having a community webinar right afterwards, and it turns out that I'm the host of both both sessions. So we have to kind of cut this really at the top of the hour, probably like a couple of minutes early okay, because otherwise we cannot start the community. The webinar.

E

Does it also someone.

A

Else host, maybe I can try. Let me just try this as long as somebody else has the host, it should be fine.

A

The question is: if the recording keeps recording, if you change the host.

A

E

Right well, the next issue is about kind of prep work for the self-healing use case, where we need to refactor the remediation service and, in this particular issue, I want to extract the feature of posting, a command to a dynaTrace problem problem into the dinery service. Right now, it's implementing the remediation service, and this is the wrong place to live and yeah. With this feature, we want to just extract. Is this functionality into the territory service? Yeah I've explained everything here, meaning which event the service should react on?

E

What to do, and here is one also one point to consider it's regarding the secrets that I use Palatine address service. They can be also staged defined, but here is a link to resource code that might be an option for you, free reusability and yeah.

A

So if you essentially don't just copy paste the code from remediation service over to dynaTrace service but reuse, the function that's already implemented in the dynaTrace service, then it should be fine.

E

All right who is up for this one.

E

Okay and very similar to what we are doing here with the dynaTrace service, where we extract feature feature from the remediation service into another, one is also implement or is also the task for the next issue, because right now, the remediation service also does the scaling action scaling the replica set by one, and this should be moved to the helm service, because the home service is the guy who can take this action and can execute the action. This should be done by the helm service, I.

D

E

A

A

Yeah, if so, I'm very sorry, yeah you're, honest you are the host. Everything should see you fight.

E

Okay, then we have one two, three or actually just one. Two issues left that are def focused, it's refactoring. The logging mechanism in the captain core services has some one time to address this. One.

D

So we can leave it in the background. We can't.

E

Believe you on.

D

E

I want to get it implemented for ordered seven, because we, this is a kind of refactoring task, and we want to address this issue, but we can leave it there and I think the same goes with the with the next issue. That is all about adding the problem ul to the payload of a problem.

E

This one we already discussed and down there I have a placeholder that reminds me to take a look into the repository of the of the of the captain, documentation and I pulled in three issues.

E

The first one is guys we're actually missing information on the captain, Docs page about sto, when someone want to figure out what we are doing with this deal, we can't provide information there and I will add section to the documentation to make it clear and to make it obvious why we are using is to you and which version we are using, and since we are planning to upgrade easier, I also need a better understanding of this one, and this goes way way. This goes along with with providing the documentation yeah.

A

We actually already have a page about helm and think that works.

E

A

Fine, if I just have a side by side with East you I.

E

This is the same way I propose to implement in.

A

E

Write the documentation here.

A

But I'm back, we cannot start the way since this meeting is running. Okay.

B

A

We have to kind of cut this meeting. Early I would hope, maybe the book it picks the issues and, besides himself or herself and yeah thanks for for this meeting today, I'm sorry that we have to cut it a little bit early. It's a good learning for the next meetings that we cannot schedule away, we're now right off the words. Otherwise, we cannot start on time now.

E

I think I think we are fine. There are just two issues left, that's not a big deal all right, then thanks a lot.

A

And just be aware, the next community meetings are, we have a change of dates, just check it out in the community calendar. Okay, thanks everyone, bye-bye.