keptn Community, 23 Jul 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keptn Community Meeting - July 23rd, 2020

Description

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

🎉 Keptn 0.7 Release 🎉
CNCF Slack Channel: Join here
Good first issues: https://goodfirstissue.dev/ or start here
Keptn mailing list / google group: https://groups.google.com/forum/#!forum/keptn
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.
Andreas:
Rename install flag keptn-api-service-type into endpoint-service-type: https://github.com/keptn/keptn/pull/2099
Handle empty creation date in get projects: https://github.com/keptn/keptn/pull/2091
Imre:
Fix error message when checking for K8s version: https://github.com/keptn/keptn/pull/2070
Johannes:
Align version check in CLI according to implementation in Bridge: https://github.com/keptn/keptn/issues/1930
Prepare Keptn 0.7.0 Documentation: https://github.com/keptn/keptn.github.io/pull/579
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
Keptn 0.7 Power Demo:
✨ Closed-loop Remediation with custom Integration - KEP 09 | SPEC 31 Please find a detailed blog post about this use case here. (Johannes)
🎉 Improved automation support with API extensions - KEP 10 (Johannes)
🔒 Hardening of Keptn (Johannes)
🌟 Continuous Delivery with Helm 3 (instead of Helm 2) (Andreas)
⭐ Removed Istio and NGINX - KEP 18: Documentation is provided here. (Andreas)
🚀 Delivery Assistant - SPEC 26 (Andreas)

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

Out all right, recording has started then uh welcome everyone to this community meeting and death meeting of the captain and um yeah like every week. Let's go uh and let's have a look on the agenda for today.

A

Well, we have a big announcement. uh Yesterday we released kept no.7 and it's now publicly available um yeah and feel free to use it. Just a um just a note. We will do today the the the death meeting today like we do it every week, but at the end we will do a short demo of the key features of the key announcements of captain 0.7.

A

Please stay tuned and yeah wait until we have done the official part of this meeting and then we will go over the new features, at least by having a quick look on each new enhancement, all right. This is now all I have to say about the cap. No.7 release and more is coming later. Today, then, a second announcement I have to share um now or since we are now listed as a sandbox project in the cncf.

A

We also got a slack channel there. This means that we will also maintain the slack channel channel and to join it. Please follow this link and then yeah subscribe to the channel and feel free to post your questions or your your thoughts about captain directly in this um cncf slack channel.

A

But still um we want to maintain or our official way of communication is our captain slag channel, but we will have a look on what's going on in the cncf one um yeah as a as we will check it daily and and let's we will keep an eye on that.

A

Then, um like every week, we also wanna. um We wanna tell you that there are good first issues to start developing captain when you want to get started with implementing a feature and don't know what is the best point to to do that feel free to go to good first issues, dot, death or use the link that is provided here as well, and then you will find a list of good first first issues that you can then start to to work on all right and dance like every week.

A

We also want to point out that uh captain has a mailing list um subscribe to this mailing list, and then we will get all the announcements directly in your inbox, and this now brings me to our next point, where we now go over the pull requests and the issues that have been implemented uh this week. We will do it um yeah one by one by one and um andreas will now start to talk about his features. He has implemented.

B

Cool. Thank you.

B

So, as johannes already mentioned, this week was uh special to captain because we have released 0.7, and this is also a reason why we do not show now a big new features today.

B

Instead, we have small prs, which we would like to discuss, which have been implemented for the new release. So my first one which I would like to show is I've renamed flag in the cabin installer and this flag.

B

When you type cabin install, you can specify the service type of the cabin api, and this was a little bit misleading because you do not only specify the service type of the api, but also of the bridge. So I really renamed the flag and the endpoint service type and yeah small changes in the code, but as this is customer facing, I would like to point this out.

B

So my second pull request here. I addressed a competiti issue, so, as you know, captain always provides upgraders from an old version to a new version and with captain 0.62 we did not had any creation date so for project services and stages and therefore the command can get projects failed, because here there was no creation date available and um I fixed this and this put request.

B

So that's basically from my site, um if you have any questions to this to pun requests, please let me know otherwise. I also see in gray in this call, I would like to hand over.

C

Okay, thank you, andreas. Oh, I'm gonna share my screen.

C

Okay, so uh I work on the bug fix uh for uh install cli go here, so it's just one liner changes so basically previously, if there is a failure in this uh call, basically for checking the kubernetes version, it will return the irrelevant error and now I'm uh just another error saying that uh the error is failed to check the google reciprocation and wrap the error given from the check tube server question.

C

So I think that's all from my end. uh I think I will hand over it to johannes.

A

All right um yeah in the introduction, I forgot to mention that this week is also a little bit different compared to the others, because a couple of folks are already on holiday on vacation, and this is also the reason why we are a smaller group today, but yeah. That's that's the way it is, and that's.

A

Okay, now my work this week was mainly focused on updating the documentation for the captain 7 release, but I also worked with andreas on getting this feature done on getting the feature done that says, to align the version check in the cli according to the implementation in the bridge, and what we did is what, when you now say, captain version, then you will get the current cli version and you will also get an information that the daily version check is currently enabled, and this means that the cli will check for new versions every day and when you want to turn off this feature just set this config, the config parameter.

A

That is called automatic version check. um This is the parameter the flag and you have to turn it to false to turn off this feature. Let's do this real quick now when I turn it off and I'm again typing in captain version, then the command is telling me that this daily version check is disabled and you.

A

I will not get informed about any new uh captain version yeah, and this is now aligned to the implementation in the bridge which all which also provides this feature, and when you want to get more information about um the daily version check feature. Please follow this link, and here you will get all the information um uh about the feature itself, how you can turn it off and turn it on.

A

All right yeah, this was the first one and then let's have a look on the documentation for captain 0.7 because it changed compared to the previous releases. And let me just first, let me go to the old documentation in the old documentation.

A

We wrote the docs in a way that we followed a use case or followed a tutorial and therefore most of its most documentation was written in the tutorial section and yeah. We provided then guidelines to onboard a service guidelines to do deployments with quality gates and so on and so on and since jurgen did a really great job on extracting this information into the tutorials page, which is now provided on captain.tutorials.sh.

A

Tutorials.Captain.Sage here you can find find all the tutorials. We also decided to not structure the documentation in this way and in 0.7 you will not find tutorials or any guidelines for doing certain things. No, we structured it now in a way to have it more focused on particular tasks like, for example, how you can operate captain how you can manage captain then also you will find documentation for the continuous delivery part of captain. Then the same will be provided or you will find documentation for the quality gates and automated operations.

A

And then, when we take a look into one of those sections, let me take the continuous delivery one, and here you will then find how to how to use about exposing deployed services, setting up a multi-stage, livery, workflow or information about the deployments with helm and so on, and now they are decoupled from use cases and documentation that stand by their own. So to say and yeah um we we of course we took it very careful or we were.

A

We were very carefully when, when writing the documentation and when checking each link provided in there, but when there is a broken link, just shoot me a slack message that you identified one and then I will tear take care of updating, broken links and just a maybe a side note on this. I think it's tonight or tomorrow night.

A

We will do a check on the documentation to identify all broken links.

A

All right, yeah feel free to to click through the documentation to get familiar where you can find information and um provide feedback. Your buyers, like.

A

All right, yeah, a lot of work was done for captain node.7 talks.

A

Then before we go on to discussing uh upcoming tasks, I want to point out that um we also we are already planning captain seven and for getting involved into the discussions that are going on around captain 0.7. Please go to this captain enhancement proposal. It's captain huntsman proposal number six.

A

This proposal has the title: the next generation of shipyard and when you go there you will find um you will find the documentation. You will find a proposal how we think that the shipyard should look like in odot 8.

A

Feel free to to ask questions directly on this pull request or to raise your concerns to bring in your ideas and to get involved into discussing the next, the next way of declaring a shipyard file- and this will be the driver of the road map for cap node.8 as we are, we see potential use cases there and we also want to implement- or we want to get to a point where we can split the captain control plane from the execution plane.

A

But this is a topic I will touch on next week to have a outlook on what's coming next, but for this week for the upcoming one, please feel free to take a look at the captain enhancement proposal and to provide your feedback.

A

A

Perfect. That brings me now to the point where we jump over to the work in progress: kanban board. Let's just go there. It's captain.

A

A

And we go over issues that have been presented today, and this was imreas implementation of the of the cube cartel version check. Thanks for showing us your pull request, then, as part of the releasing process, we updated the documentation. This was joint work amongst all team members, and I showed you that andreas and I we aligned the version checking the cli according to the implementation in the bridge.

A

Since florian is not here and neither is uh airmen, I will keep those in the backlog or I will keep those issues in the ready for review section, and we will ask these two gentlemen to sh to explain their features uh next time.

A

Then, in the progress column we have the release 0.7. Let me move it over to the ready for review as yeah. It's done. It's done done and no further actions.

A

Required, okay, today I had a discussion or I had a talk with uh airmen about his two issues there. um He will provide a pull request by tomorrow, where he does some refactoring on the bridge, and it also creates helper functions for utility utility functions, and he also wants to provide a pull request by tomorrow, and this is just a reminder that we don't forget to archive the captain.

A

At the geometry extended service, repo.

A

All right, then, let's go over to assigned and committed issues. The first two ones are also assigned to airmen.

A

He wants to have a look on the bug that appears when, when there is a when there is a evaluation that did not work or no a test, when there is a test that did not finish correctly or in an unstable, um with an unstable or unknown result, then the heat map is broken, and this is a kind of corner case and amid wants to investigate that and wants to provide a bug fix there, and he also wants to do some refactoring yeah still ongoing, but um he's taking a look on the code and inred.

A

This issue is assigned to you. Actually you assigned yourself to the issue. Thank you very much for doing that. Are you still okay working on that issue,.

C

uh Yeah, actually, the previous installation of captain didn't work on my laptop, so I will try to install the latest version. The 7.0 and we'll start working on this one.

A

Perfect, thank you very much, um maybe just a quick reminder for everyone who joined today. uh This issue is about implementing or changing the response code of of an endpoint, because when you now uh do or get on the events endpoint, I think it is yeah when you do a get on the event end point, then the response, the response is telling you that there is an internet there, server error, but this is not true. It's just the case that the resource has not been found and we want to change it from 500 to 404.

A

Thank you very much. Then we pulled in to um yeah a couple of issues we want to work on this week. The first one is that we want to formulate the spec for shipyard 0.2.

A

All right yeah, as I already mentioned, we have an ongoing discussion in the captain enhancement proposal number six and in this proposal we have a new specification for the shipyard file and this specification should be implemented in the captain utils package so that we can use this struct for for implementation.

A

um And I've before assigning this issue to someone, I want to also open that one, because this is also a spec related issue where we oops sorry.

A

Maybe, but we want to work on the events and in the future, each event will be a combination of triggered, started, status, changed and finished. This means that, for example, a testing event is triggered by a test.triggered.

A

Then the service that received this event will send out a test dot started in case there is a change or something uh changed during the process of of executing the the job the service can respond with the status thought changed and, finally, when the job is done, he has to respond with us dot finished, and this is also a spec related task, where we want to define those events for a typical deployment and auto remediation tasks.

A

All right who wants to work on which item.

A

B

Take the second.

A

One you take the second one. Thank you very much.

A

And then I will take care of providing the specification for their shipyard auto too.

A

A

All right cool, then I have here a good first issue that says there that we can or should improve the cli with a better error message, because when we have um an environment where captain the captain cli cannot communicate to a captain cluster, for example, because there is no internet connection or the the api of captain has changed, then it should be more.

A

Then it should be possible to get better and better messages, because, right now it's just telling you that, for example, adding a resource could not be processed yeah by this by this error message, but in the future it would be cool to get it to get at least on some guidance why the command did not work or why or how you can fix it.

A

It's, um I think, it's a good first issue, since it's mainly related to the cli and checking the communication and the error message from the api endpoint um and when everyone, when someone wants to work on that, um yeah feel free to to take it and to to work on on getting better error messages.

A

All right, then, I still have this issue in there. It's about the heat map um yeah, but since airmen is, is already busy with refactoring and getting the bug fix for the heat map.

A

I think we will keep that one in the backlog.

A

A

Any thoughts, any idea or any concerns about the work for the next week.

A

No, then we discussed work in progress. We did go over them, then the work for the next next week and we did a planning- that's great. Let me just close that one.

A

Cool then we are now at the point where andreas and I are giving you a brief demo of the main use cases that were implemented in 0.7.

A

We have one two, three, four, five, six key announcements, um yeah.

C

A

Announce uh the first one I will now demo. You is the closed loop remediation and also now, with the possibility to plug in custom, integrations the underlaying us captain enhancement proposal and spec change is referenced here, but yeah, but it's just for documentation and not required anymore.

A

All right now, let's show you what has been implemented in regard to this use case and for doing so.

A

Let me first go to to actually to the repository to the repository that contains all the the runtime configuration and also the configuration for the remediation task for a particular project and service, and in this case I'm now using the project sockshop, which has two services onboarded.

A

Let's, let's go to the root folder and in this project we now in this repo. We can see that we have two services: the cards in the cart db service and when I click on the cards service.

A

Here we can see that this card service has a remediation.yaml and also a service level objective.yaml.

A

And when I open this remediation yaml, this is different compared to uh which were you which, compared to version to the version we used in 0.62, because what you can now do is you can now specify remediations in there for certain problem types and the problem type is mapped to the title of the problem that is coming in, for example, and in this example, I have two problem types.

A

One is for response time, degradation and the other one is for a failure rate increase, and when a problem comes in with one of these two problem types, then the action which is configured below will be executed and the action is, in this case, a scaling.

A

This means that the pot will be increased by by one and in case of a failure rate increase, it's toggling, a feature flag.

A

An action itself has an name, a description and in the values property you can specify values you want to send to the action provider.

A

The action provider is, then the service that takes care of executing the action and and responding back with the action.finished.

A

Now, with this configuration added to the service in a particular stage, we are ready or we are on a safe side. Whenever a problem occurs uh with the type response, time, degradation or failure rate increase, then captain will take care of executing the corresponding action.

A

And now, let's show you or let's do this- based on an example and therefore I'm telling you that currently I'm running a load test, this load test is sending requests to my service and this we can verify by taking a look into dynadress. uh Here, I'm just going to transaction services.

A

I have the sock shop, production, environment selected and when I open the service, which is this one, then we can see that this service is hitting or receives a lot of requests. These requests are coming from my load test.

A

What I'm doing now is that I turn on a feature flag and the story behind that feature flag is that we want to enable a promotion campaign which is implemented in the service that is currently getting the requests, but the promotion campaign runs into a failure rate or runs into a failure, and then we will see that a couple of failures are occurring.

A

Let me just refresh.

A

Here we can already see that two red bars are now showing up. These two red bars are indicators that failures are occurring and it takes now a while for dynadress to open a problem and then to send out the problem event to captain, and I have already prepared a way to uh to send out this event in a faster and what I'm just doing is I'm sending. Now. um Let me just show you the event itself, I'm just sending handcrafted problem.

A

We went to captain and this will then be the trigger for doing the remediation workflow and for sending this problem event. I will do captain send event minus f and I will select the failure rate problem.json, which is a cloud event, and this cloud event is now sent to captain.

B

Can I quickly interrupt you and ask questions so um what you're doing is you're playing the inner trace and the reason for this is you do not want to wait until then androids detected the problem and identified through root cause right right.

A

Yeah- okay, I'm just um that's true- I'm now playing dynadress by just manually, sending an event to captain and to inform captain that there is a problem: okay, because it takes about roughly uh up 10 minutes to for dynadress to identify the problem, but in captain we can now see. Let me just refresh the page to be sure that I'm showing you the the right problem- and we can now see that the problem came in it was at uh 534, and what now happened is that the problem was received by captain.

A

Then captain trigger the remediation, because if captain found the remediation file, I showed you then status changed event. This is sent out. This is uh this: has some informative character, but more important is that an action was triggered.

A

Actually, the action which was yeah specified for a failure rate increase and then the unleashed service, which is deployed and took care of executing the action, and he also finished the action and what he did for us. He turned or he toggled this feature flag. I just refreshed the page. I did not toggle it by hand. This was done by by the unleash service and it turned off the feature, and this means that now we should not receive failures anymore.

A

Let me just refresh the page again.

A

And it's true, yet the feature flag was toggled, the feature was turned off and now the failure rate went down to zero.

A

And I'm not finished with my story because um after an action was executed, captain also takes care of validating the execution, and this means that captain now will wait for 10 minutes. You can configure this depending on your needs, but after these 10 minutes an evaluation of the quality gate will will be triggered, and this then ensures that the action worked or did not work.

A

And again we have to wait now 10 minutes, but we can also take a look on a workflow execution that was executed before before and, let's just take that one, it's the same, one problem came in, the remediation was triggered, the action was executed and after 10 minutes, you can see it by comparing these two time stamps.

A

We started the evaluation, the evaluation fetched the data from the monitoring solution and then did check on the quality or did a check on the service level objectives, and in this case the evaluation failed because no data was received by by the sli provider, but it is at least a validation, whether the action worked or did not work.

A

And when this evaluation is done and when no further actions are configured then also the remediation is marked as finished and the workflow is is is done and yeah. That's then,.

A

All right any questions I mean this was now a very, very brief overview over the use case over the new features, but when you want to learn more, I can also point you to a blog post that I wrote just follow this link. It's a blog post or medium, and this blog post explains how you can set up your remediation workflows on how you can provide your action providers for for captain when there are no further questions. I will go on with talking a little bit about the improved api extensions.

A

This this enhancements were driven by a captain huntsman proposal from christian. um He proposed that he want to get more apis or end points for doing automation around captain and what captain 7 now offers is a really powerful api.

A

Let me just go to swagger. Let me just go to the api of captain and there you have now the possibility to select between the api service.

A

Then the configuration service of captain and then also uh you can also select the mongodb data store. Let me just tell you what the api service is. This service is responsible for doing the authentication.

A

Then you can also get metadata from your captain installation. By this end point you can post cloud events to the event endpoint and you can also do a post on on the project and on the service endpoint. This means that these two endpoints they will create a project and the service.

A

Then, from a configuration service point of view, you have the possibility to also get all the projects that you are maintaining with captain right now, and you can also fetch the resources that you uploaded or yeah that you uploaded.

A

And based on those enhancements, christian also helped us to enhance the captain cli, and what you can now do is you can now type in captain get project.

A

And you will get a list of all the projects that are currently maintained by captain. Then you can also do a captain get services.

A

And you can also do a captain get services, minus minus project equals sockshop, and then you will get just the services that are onboarded to the to the structure project.

A

This now uh helps everyone to to deeper to understand, what's currently onboarded what's available and that's really cool work that also christian helped us to implement.

A

And let's just execute one endpoint together: let's do, for example, the the project end point. I hope that I'm authenticated. No, I'm not.

A

Because this end point also gives you now a lot of information about the project. Try it out.

A

Let's execute this endpoint and you can see when you execute um slash, get project we'll do running this this or x um doing a curl against this end point, then you get a list of all projects: the creation date, the remote url of this project and also the git user that is used for this project and the name and then also a list of old stages and in each stage the services that are deployed and for each each service.

A

You also get the events that were sent out for this service, but just that the last one, not the entire list. You will just get the last event, for example, for a deployment finished for an evaluation done and so on.

A

All right, then, I'm going to my next issue, which is our next announcement, which is hardening of captain.

A

We invested a lot of work into understanding the permissions and getting a clear set of permissions for each service, and for that I want to show you when you now go to captain then to the installer.

A

You will find no installer job there anymore. There are now just manifests, which is a great effort.

A

Then we go to the templates oops. Sorry, we go to the charts.

A

Of course, then we take a look into the control plane of captain and in there we find the role based access control, and here we now clear, clearly clearly define which service accounts are required, then also which roles we want to use or want to configure want to use for captain, and they all have a clear set of permissions now and then we also assign those roles to to the services by using the role binding, and this now um is helps us to get better understanding of what's going on and also which permissions are required by which service.

A

All right, yeah.

A

Actually, there is nothing to show um it is applied to your kubernetes cluster, but now you can be more or now we are on a safer side and also you can be on a safer side of understanding what is executed and which permissions are required.

A

All right, then I will go over. Oh there's now a question coming in. Does it still try to put the installer in default? Namespace? No, and there is no installer job anymore.

A

We now have just one helm chart and the captain cli um a captain. Cli applies this helm chart to your kubernetes cluster, but there is no job that gets that gets, triggered and executed.

A

I hope I think this answers the question. Yeah it disappeared. Thanks then andreas can you take over sure.

B

So I would like to start with two rather technical highlights. We've made and the first one is that we are now using helm version 3 for continuous delivery. What does this mean? This means we do not have to to install helm anymore.

B

um When you have a look at your parts in the cube system, namespace.

B

Okay, you will not find till anymore, so.

B

So this is a major improvement, because um this re moves for us in external dependency and only our health service requires uh the dependency to handle three, but not captain.

B

So and if you're feminine there miss captain with helen. You know that this hidden releases are now stored as secrets and respective namespace, and I do have hand releases in.

B

Sockshop- and here you can find, of course, the releases which are available.

B

Of course, um we do not leave you alone, um with the upgrade from heaven version 2 to head investigation 3, and you can find all those details in our documentation go to captain.sh docs then operate can and then upgrade kit and.

B

It seems that my internet is a little bit.

B

B

No, we got it um so we are providing an upgrader which allows you to upgrade um captain version 6.2 to 0.7, and you can now do this by also upgrading your hand, charts or by not upgrading your handshake. So this is without the upgrade of the handshards, and this is with upgrade of handshots. If you decide not to do it with captain, of course, you have to do it in manually and you can find the respective descriptions and commands here.

B

So this was my first technical highlight the second one is um we removed the insulation of istio and engine x from our installation process and all details can be again found in the, but we just got the question when we still have an insular job, and the answer is no. We do not need those job anymore because we do not install istio no engine x by default anymore.

B

So this gives you a lot of freedom now how to expose cabin, and I did and.

B

uh And visualization how you can now expose cabin and basically, we identified four options up to one is using an ingress and for those of course you need an ingress controller. This can be easter engine x traffic and then you install captain and apply your ingress object inside canon. The second option is to use load balancer, and this is exactly the option I've chosen in my installation, okay,.

B

You can see.

B

That I've used load balancer, which gives me an external ip address, which I can then use in order to access the api. Now, let's use this.

B

Ip address and you can see, I can access my api. If I do not specify the sub pass ifi, I get the bridge and its dashboard.

B

The third option would be to use notepod so um using the ip address of a node and exposing one port, which can then be mapped to the api gateway nginx, which is a service of captain and again you can access captain via this node and finally, of course, you can always do a port forward and using cube ctl minus n port forward in order to access cabin, and this would now then map the endpoint to localhost.

B

Using this cat endpoint, you can do the authentication of captain. So in my case, I'm already authenticated authenticated against this cluster and you're good to go. So this was the third technical on the second technical highlight. And finally, I would like to present you the delivery assistant, which is a really really nice feature when you're using cabin in production.

B

So what you can see here is a shipyard file and you probably know most of the shipyard file except the definition of the of the approval strategy. So what I did here is I defined an approval strategy, and this this approved strategy requires a manual approval before anything gets deployed into my second stage in the staging. So before I do this deployment step, I need an approval which uh user has to do and we will afterwards see how to do this. Approval now, you're, possibly wondering why we do have to differentiate between past and warning.

B

So this this allows you to to use the emulation cell and result of the last stage in order to decide whether an approval is needed or not. For example, you can automatically pass if the here I have an example.

B

If the emulation was almost a passing result and then the artifact automatically gets gets deployed into the next stage and, for example, do if you have a warning evaluation, you can do this manually, so what I've prepared now for you is I a few minutes ago I sent two new um artifacts for our card service in version two and three two to the into our cabin.

B

However, you can also see that captain did not do the deployment in staging. Instead, it requests for an approval, so an approval triggered event was sent, and this is also true for my second service. When I now go to the.

B

Cli, I can now request exactly those open approval events, and you can do this with captain get event approval and I'm in the project.

B

Workshop and I would like to see the approvals for the stage staging.

B

And you can see here two open approval events, so it's I can see that um there is an open approval for the image version, 11.2 and also from for 11.3.

B

Of course, you can then also confirm an approval using the cli and therefore you can use the command captains and event approval finished and captain again shows you two options. You can either approve um 11.2 or 11.3, and um actually you could see here the emulation results, but in my case in my installation I did not do any evaluations.

B

So if I would like now would type one and press enter, I would have approved version number two and this would get promoted to staging.

B

But I don't want to to do this now, because I would like to also show it in the bridge, so you can also use captain's bridge, which is the ui of captain in order to see open approvals, and you can see here a blue sign which shows me hey. There are open approvals and again I can see um cards in version 11.2 and 11.3, and now I can use the bridge in order to promote this artifact and I would like to have version 3 in staging.

B

Therefore, I click here on 11.3 and go back into our service screen and you can see here, approve finished was send it and now the configuration change kicks in and sends this approval finished.

B

So I do have to reload it. Yeah approval, finished percent and configuration was changed which triggers new deployment now in staging.

B

So that's, basically the delivery assistant, and this really nice finishes our presentation. So if you have any questions, please let us know we have three minutes left.

A

Yes, there was one question coming in while you were talking, and the question is where istio is still needed before installing captain.

B

Emery, that's a really cool question. So, in order to use captain, you do not need istio or nginx at all, but for doing blue green deployments. uh In the background we are using virtual services and those can only be used if you have issue installed. So um the there is no clear. So the answer is for captain. You don't need istio, but in order to do the complete, continuous delivery workflow in order to do green deployments, you need istio.

B

B

So um does anyone have a feedback on? One of those highlights we've presented. So is something like the delivery assistant which I've shown useful for you. So my question would be in your company. Are you using a manual approval before promoting to production, or is this everything.

B

B

Okay- uh and there was actually a second question from emre, so uh he is asking whether it is possible to specify the the image version on the slide when doing the approval. So this is um not possible, and this has a good reason, because captain um follows the approach to have never breaking pipelines, so each and every artifact has to pass the quality gate before, and so um you cannot simply overwrite an image using using this approval and this approval um finished event.

A

There's also a check implemented. This means that when you send an approval finish to captain, then cap will check whether there is an open approval for that finished event, because in case this check is not there. Someone could hijack a captain and could pull in an image that is not validated in the previous stage and to avoid this security problem, we decided to to not provide um or to not specify the image when you send back an approval, finished.

B

But if you're really sure what you're doing you can always override, for example, the image in production, so but this is not possible with the delivery assistant. Basically, you have to send a new artifact event for production. So then you can simply replace the image in production.

B

This can be useful if you have any security problems and you need a really quick fix in.

B

B

Right and that's cool so um for all of you, um emre um says thank you for the for the answers and yeah. That's it.

A

Okay, yeah, since we are on the top of the hour. uh Thank you for joining this captain dev meeting.

A

um If you have questions to the 0.7 release, feel free to to post that in on the slack channel, we are happy to to get your feedback to take care of your concerns and also to to address your questions there. Otherwise have a nice week. We will see you everyone next week on thursday at 5 pm and what I have to say is have a nice evening then see you.

A