Project Harbor Community, 14 Jun 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: CNCF Harbor's Community Zoom Meeting - June 14, 2023

Description

CNCF Harbor's Community Zoom Meeting

A

B

Should be recording already hello, everyone, today's June, the 14th and that's official community meeting for project Harbor as such, please follow the code of conduct and just be nice to each other.

B

um With that uh we can kick off. I'm gonna share the meeting minutes link into the chat. So if someone feels like adding yourself there or adding your topics uh feel free to do so, I'm gonna share that thing. So we can see what we have.

A

B

I see some folks added themselves, new names I see. So, um let's start with the updates, and then we can switch to discussion and then in the end, if someone feels adventurous, can you introduce yourself uh to the call so uh Thomas? Do we have Thomas Nicole?

B

Yes, yes, all right.

B

Do you want me to still shrink or you just want to go over the the last Meetup that we had.

C

Sorry, what do.

B

You ask me now: you've you've added uh in the updates part, the hardware operator uh community meeting that we had so I. Suppose you you want to discuss something there or oh.

C

I didn't put it put it.

B

Or I did or I did, okay.

A

B

Our last meeting was two two weeks ago, so many things happened since then. So yeah.

D

B

I suppose, then, in that case, I added this one yeah uh just to remind you, everyone that that we, we have the community meetings for the harbor operator already going.

B

So if you feel like joining it, and you can check the recordings from this one, uh if you're interested what kind of effort uh we do there and what kind of discussions we have, uh we also have uh Community meetings notes for the harbor freighter itself, the same format as we have right.

D

B

This one, so you can, you can check them out as well.

B

I'm, sorry Thomas, uh sorry, it was like two two weeks ago, no problem so I I should be adding my name when I add topics for this question.

A

B

Your name yeah, uh hey, my camera- is off um all righty, so all right, I'm sure the next one I'm not sure if I read it but about the logging.

B

Anyone wants to take this one.

B

I think we we added this one, because uh we had it in the um said in the world map and it's just like for for the upcoming release, and there was some some questions around this one. Anyone from the maintenance can can come into this one.

B

Did you manage to take a look at this one? It's a long one. So if there's nobody who can.

E

Come into this one or.

A

B

Come into this one, we can switch to the next one.

C

So is it possible for on that we can have a proposal for this I mean uh it seems like uh we introduced a new feature and.

C

B

I'm, not sure I understand. um Can you rephrase please.

C

I mean not: uh do we need a feature proposal for the uh Json logger? um It's it seems like uh it is a new feature for Harbor, so we yeah so I feel like that. We should have a proposal for this.

B

It fixes an issue which introduces a new, a new set of features.

C

B

Okay, so who's that it's investing one okay, um either me or you uh we'll drop a message. Let's see, okay,.

B

Yeah we've exchanged some some stuff, but yeah there's no formal uh proposal for this one but uh yeah. If you think that that thing introduces too much changes, yeah.

C

B

We should we should go uh with proposal. um Otherwise, if it's some minor changes, but the amount of code that is changed, I, suppose it's, it's not just.

C

B

All right, um I won't close that thing and.

B

And we'll address that too, to the author of that PR. So we can, we can have it all right.

B

Thank you, everyone for adding yourselves there. um Stefan I, see you added something, but.

B

F

Is it so it should be any discussion topics. Yes, the last one.

B

Oh okay: uh let's first go one by one yeah.

G

B

F

B

Page, okay, yeah sure, uh okay, so Andrea uh proposal issue draft the do.

A

B

Take over and share that with us or should I.

E

uh Hi yeah yeah sure can I can share yeah.

B

Okay, stop sharing.

E

You see my screen.

B

E

Okay, great um thanks everyone for having me, um so my name is Andrea frichtali I'm, the co-founder and maintainer of a project called City events. um That's a project hosted by The Continuous delivery Foundation, which is a sister Foundation to the cncf um and yeah.

E

So I I noticed that um back in Amsterdam, uh Harbor announced Cloud event, support um and I thought that's a great feature to have for a registry, and- um and it resonates with me, because in the NCD events, um what we are trying to do is to bring interoperability in the continuous delivery space or software life cycle. If you will, that includes all the components from your CI CD, but also artifact Registries and deployment tools, etc, etc.

E

And the idea is that we we to provide like a common language for all the tools to communicate what they're doing so that you can use this information either to build something like an audit Trail for your artifacts, but also to do things like visualize your workflow, across different tools and help with interoperability, so that you have different Tools in your organization that you get consistent events from all of them.

E

It makes it easier and also having this common language enables to have things like um even driven workflows, um and one scenario that I was chatting about with the Argo folks is that the image updated image updater feature that they have um so basically uh today. Our argosity is a continuous delivery tool for githubs and they listen to a container registry and they look for updates on a specific container image.

E

And if there is a new tag, release for instance, then they will make a PR in gate and that will trigger a deployment and so they're interested in having something like using a different type of workflow. That is more even driven so that you they would get notified if there is a new.

E

um If a new tag for assistance for any imagine, they could react to that directly.

E

And so that's why I'm here I'm reaching out to to the Harbor community I think because I wanted to propose for Harbor to support City events so that, um within the cloud events format that already exists as an alternative, they could produce CD events as well. That would bring this interoperability feature to Harbor, so I I'm new to the community, so I hope I'm doing things right.

E

um I created an issue to get started and to provide some context um and yeah here is some information that I've been talking about. There is this picture that gives you an idea what we are trying to do with City events, um so I have all different tools for using City events. You can get all this information in an Evidence store and build nice things out of it, and you can also use these events to to trigger your workflows. Like I was saying and um I started a proposal.

E

It's a I can document for now, but I'd be happy to. If there is interest in these that'd be happy to make this as a formal propose of following the Harbor community process and in this proposal for now, I put some context about City events.

E

The motivation as I was saying to provide you interoperability across registry and tools. In terms of use cases, I've introduced the Argo CD use case for githubs that I mentioned. There was a comment from valim about flux, CD, and um so we started the conversation with the Flag City as.

B

E

um But we are not. uh That is that far as with Argo, so Argo is interested in implementing City events.

E

uh Possibly uh labeling artifact could be also uh a use case for this.

E

And yeah so in terms of uh design, so there is one event that City Event specification covers today. That is meant for registry artifact registry to to produce, which is the artifact published event. This is an event that is produced anytime. There is a new version or a new artifact published into the into the registry, but we could add more um on the city event.

E

Side and I would be happy to to drive that work and I already opened an issue on City Event side to end things more like artifact pulled or artifact labeled uh I've seen you have an artifact deleted event. Today there is an artifact scan that we might add as well, um and so this would be the primary the first goal for this design.

E

There is also a possibility to support consuming events um and then so the the idea there would be that if there is an event coming into Harbor, a CD, uh for instance, from a testing tool that executed a scan on the artifact or perform some extra testing, then that could be used by our book to label the artifact. Accordingly, that's um so consuming event could be a bit more challenging, um so there are a couple of options.

E

How things can be done and my general recommendation would be to start with the producing side, and also there is a more clear um use case there and then possibly continue on the consuming event side of things um yeah, so I didn't want to take too much time. So this is kind of the basic information I prepared for today and I would open it for a few questions that you may have.

B

First of all, thank you for the interest and willing to to work on this. One um and second uh I saw on the technical implementation like in the very end. You need some input from the community.

B

If you can formalize that thing, what exactly you need, um so we can address that to you, the maintenance team uh to to give you input on this one.

E

um Yeah, absolutely so I'm very new to the harbor in terms of like I've not contributed to Harbor code base before uh so I'm new to to the community and the processes and so forth. So um it will be challenging for me to to judge what is the the best way to to introduce City events within Arbor.

E

um I could probably imagine that, since there are some notification formats available today, City events could be an alternative one uh but yeah. So there are a lot of details and design discussions to be um well, not a lot, but a few to be discussed about like how this will be configured, how it is exposed to users and so forth. Right so I, don't know there are what kind of consideration needs to be taken um but yeah. So um what.

A

Are the next step from here yeah.

B

Go ahead: um okay, someone else wants to yeah go ahead.

C

uh Yeah yeah, uh first of all, um thank you for uh uh for the proposal. So um so uh can we uh could you have to open the new proposal in the Harvard Community repository so that we can uh um discuss a more details based on that proposal and yeah yeah I? Thank them, and second um um I can give you some more more background uh regarding to the uh event integration.

C

um So by uh by introducing the support of cloud events, we did some fundamental uh reflector on the on the hamburger base and, after that, um it's quite easy for users to introduce a new kind of format. For events like someone is working uh um from someone from Community is working to introduce the um teams.

C

uh You went on into Harbor and you can refer to that PR to get more details, but but basically uh it's not quite difficult for someone and who not familiar with Hardware code base to introduce a new kind of format.

C

uh I mean the the event payload and, and you can connect to a a chain to get more help and he did a refactor on the the code base on so.

C

B

C

B

So, to sum up, um if you can translate that thing to uh um Community repository proposal, so people can start adding ideas uh it. It will be a bit more messy uh I understand your initial idea to do you think I can do it, it's a bit more clear. When then, you create the actual proposal, um but then we are losing all the history of decisions uh in the in the proposal I'm going to link you some some uh some peers that were implemented.

B

uh What Ian said with the with the recent changes and some some new implementation so can guide you how to navigate through the source code and how you can Implement that thing.

E

Great yeah sounds great thanks: yeah I started with an act and D because I wasn't sure if it was okay to create a proposal right away, and it is to present it first but yeah I'm happy to to create a PR, and then um we can discuss in terms of uh proposals, I mean for the producing versus consuming events.

E

In my view, those would be maybe two different features that makes sense to then create one proposal only for the producing side, and then we can discuss whether to want to have a different one for the consuming. Or would you prefer a single one that is implemented over time.

A

H

Cover consumes the events from outside.

A

H

Yes, so uh if power receives the events from outside, uh what how I can do for Next Step.

H

I I mean I, mean I mean receive the events for for what action.

E

Yes, exactly so that that's um not uh 100 clear to me, so there is one possible use case. um We do have like test events, for instance. So if you could you wanted to delegate certain logic to external testing platform, you could be notified that a test suit was completely reset and tests were executed, then add a label to to an artifact uh as a reaction to that, um but.

C

That's yeah yeah go ahead.

E

C

That's that's it I mean it's just yes, I I prefer that we can step separate this into two proposal, because that the currently on Hardware only produce event and do not consume human lack, and, like you mentioned in the.

A

C

Of a scenario, so we only set up on the events to the other like cicd system. So then they can. They can compose their pipeline by consuming the events that sent out by Harbor, but currently Harvard does not receive any event from any uh outside system and do any action in The Harborside. So I've I personally agree that we can step it. These two uh category into two proposals. One is introduced a CD humans, the other one is harder to consume humans from outside that okay, yeah.

E

Perfect makes sense, and it could well be that um I'm in the community perspective is that we actually don't need to consume events at all. So we stop at the first proposal and but that's it I, don't think it's common for for other for container relatives to consume events, uh to be honest, but.

A

E

Wanted to provide both use cases for completeness and see what people thought about them, um but yeah so I'll start them creating a proposal for the producing events, um and we can maybe create an issue about consuming events to see. Yes, if there is interest in this at all, um and if there is, we can maybe discuss how you could just like.

C

So our for the consuming events, you can either open a issue or discussion in the harbor repository so that someone from Community can add their thoughts or ideas. So then we can based on their thought, to continue our discuss.

C

B

A

B

Thank you very much. Anyone else want to comment on this. One.

B

All right, thank you, Andrea. um By the way, one last question um I saw you from IBM. The big blue uh is that City events thing parts of the IBM or that's a side project of yours that you're working on and I mean.

B

Is it related to your work at IBM, I'm, just trying to figure out who's who and how? How that thing works and.

E

It is also related to to my work as sipm, so there is interest within IBM to to having a standardized um I mean we we do offer like uh CD services on the cloud, and so we have like this kind of pipeline Runners based on tecton and we use artifactory also at IBM. We use a different set of tools and it makes sense so we use cloud events today to collect events across many platforms uh but yeah.

E

So there is interest in having a common format there, and we have interests also from other companies like web Fidelity. They actually they're the first adopter. They produce the Jenkins plugin for City events. You have interest from Apple Apple they're, contributing to the City events, because their devops team they're interested in this kind of tools, um yeah. Other companies involved like um SAS Ericsson, um so they're looking to have.

A

E

Kind of centralization.

B

The the other part of my question is, but you already answer that um if we go and Implement that thing and accept it into Harbor, we are also looking for like a long-term engagement with this stuff and supporting it, uh because it's I I believe it's not like a single shot kind of effort, but we need more like engagement, long term, um so yeah I got I got the answer from the rest of the thing that you just explained. So thank you.

A

B

E

So thanks for having me again yeah.

B

Thank you. Thank you for joining our community. Okay. Next, one on the agenda, we have Florian from cyber watch. uh Welcome again a new uh Community member.

A

Can you introduce yourself.

B

And and the whole story behind your uh joining us daily.

G

Hi everyone I'm happy to be here today with you, um I'm Florian, the CTO of cyber watch. Cyberwatch is a cyber security software company, founded in 2015.

G

And uh today, I show you the integration with the vulnerability, our vulnerability software with Arbor, it's pretty cool because uh on the bus we just have one-sided integration with with pulling some images from Arbor, but today we can show our in our CD information in an in the album interface and I. Think it's very good.

G

um Let me show you some.

D

G

Do you see my screen yeah.

B

G

Just to sum up very quickly: uh cyberologists vulnerability, management, software, our competitor, our qualities, rapid 7, untenable, if you know them.

G

Or is a all overall application or tool is our deploying Docker? So we like this technology, yeah very much. uh We are not specialized on container analysis because uh we we come from the generic infrastructure management, so we are best. We manage desktop server, Network device, industrial device like PLC and something like that and the contouring image is just one part of our scope and I: don't do the pre, the commercial part I prefer to show you directly uh the demonstration, so our tool is very easy to understand.

G

We have an inventory with all your assets on the on your computer information system. uh We can get uh rapidly list of information of something like a VMware, uh Visa or I. Do previous integration to scan all your VM, but also otherwise history like cardboard.

G

um If we target only the docker image scan for it's very easy, you can add an image by name, you have a view to show all the vulnerabilities or technology in style in the docker image.

G

It's very good, it's very good for scissor or something like that and some people in the in the company, but not for devops team. It's better to have the CV information on a tool for the for the devops team, so Harbor is great for that I just had we have implemented the three RP roads to integrate with rubber. If I add when you a b key demon.

D

D

G

On the Zara side, we have a hardware instance, you, you are more familiar with this and it's very easy to integrate. I like our very simple to to interact with him.

A

G

Thank you very much.

D

G

Just copy the credential.

D

G

So I have the integration, interrogation Services, you call them. So it's a cyber white software, the last version and after that, it's pretty cool to just choose some images and scan them. I have clone some registry two months ago. So.

G

I just run the scan, but okay queuing. If I see you in the result in the Cyber watch software, okay, Docker images- we see that we have the six images add on the list here. We need to wait on 10th 30 seconds to run them and we can show the results in your arrow instance.

G

It's already image. If you see this one, for example, on the bar on the end of the page, you have all the CV. We found all the security issue on the, why we are the problem, so the current package version are not up to date. So it's a Debian days image. You see that we miss one batch and we can show you all the list of the all the information of the CV.

G

If you want more details, you can come back with your cyber watch instance and you have the detail uh advisory of the of the ceiling and uh on the back, you have also the list of your wallet it has set, and you can see that you have the variable images we scan and inside the vulnerabilities of cyber watch. You have the same information as we found in a harbor, so in fact, for this image we have three patch missing the there are not the last version of radius.

G

So it's a shame for for the radius image, because we have five CD on the image scan and we have also one patch on the Libre so for TLS.

D

G

um I, don't know if you have any question this time.

I

I have a question. Thank you for real. um What data are you using for the vulnerability scan? What is debate? What is it data for your scans.

G

Okay, uh so in one side we pull the image on the over scanner and we run some bash or shell script inside to extract all the uh all the list of Technology inside the packet, the docker packet, every Debian Deb, or a very software like ready, server installed manually on the robot side, we call all information on the internet. If you see this CV in particular, we show you that we call information from the list and the nvd on the information come back from every distribution, Linux distribution like Reddit Debian, or something like that.

G

We also fall over vendor like Microsoft, Microsoft Oracle, and something like that, and we uh we combine all the information to show you the.

D

G

uh Cd, you are in your image.

G

You are mute because.

C

um Hi I I have a question regarding to the cvss3 score, and uh so uh you start imagine that you can bend this data um into the Cyber watch. So then you do a mask based on the The Source. So then provide a c vss3 support or you just use the original data like um ID.

G

Yeah, we use the original data from the nvd, so it's the same score. We found in the NBD information.

C

G

Okay, don't override score, we overwrite score only when the nvd don't have the information on the vendor uh say that the score is. uh This is a is. It says the score, because we don't uh we don't want to be.

G

An intimate we are just an intimate, enter material to the to uh and don't want to be uh uh active to show to.

G

To say the score, because it's difficult to categorize the ndd of the CVSs.

G

Criticality of vulnerabilities, so we prefer to show the the exact information from the handy.

C

Okay, so so you have uh Implement uh adapter for cyber watch based on their platform scanners back different.

A

C

Hardware, oh yeah, okay, yeah! Thank you. So edit open source on I mean the adapter.

G

No, we, the adapter, is native on the application, so it's not open source and you can use the Cyber watch uh analyzer open source. It's a paid software I'm! Sorry about that! But, okay, okay,.

C

Got it got it um yeah? Thank you. Thank you.

B

I have one um in the terms of Open, Source and really kind of Spirit. uh Do you offer some kind of pre like a open core kind of subscription, so people can leverage uh server watch from Harbor the same way as we do for 3v, for example,.

G

We have trial, but not uh it's real or proof of concept for three or two three months. Something like that. But not we have not reached here uh subscription.

B

Sure I'm just curious if someone asks, because sometimes people ask about this kind of Integrations, for example, with sneak and some others and yeah, for example. Sneak also doesn't have like the three integration to like the same way as 3v is doing it just asking for the records to be honest and out of curiosity, no nothing else, yeah all right.

C

um um Sorry I have another question: um did: did you find any uh issue on the scanner Spike during the implementation on your adapter and, like I, mean some something is not a reasonable, a lot different by the spank like request or response or.

G

We have no issue to implement the the adapter with uh with your Wiki, it's very easy on the on our workfinder with twin to to ingest data.

C

I'm singing this because uh we want to do a um implementation or or enhancement on this pack and to provide a master minorities to introduce more capability for multiple scanner like we can support, s-bomb or licensed scanning yeah yeah. The things like that so um so uh I would like to see. uh If someone can give us some feedback so then we can based on this feedback to to make better a design yeah. So if you want some problem, just let us know. Thank you. Okay, thank.

G

You I asked my developer team if but uh I don't hear any issue on this part.

C

B

Thank you, foreign. Anyone else. Any questions about that.

A

Yeah I know it's Thomas, uh just one question: did you do you have some Benchmark comparison results with a Your solution against I? Don't try this, for example,.

G

A

Do you cover the CV scan.

G

In fact, we don't have, uh we don't make a comparison with trivi, because it's not one of a competitors directly I have made some comparison with sneak because it's more customer challenges for us, but uh in fact the result uh as pretty the same.

G

The main difference is that we are more focused on series, our solution, so we don't.

G

For, for example, for Debian packages, we don't see the any CVS are not. They are not patchable. So it's when our side effect, because we don't want to lose time for our customer with unsolving problems, and this in our site. For us, can we do with the CPE from the NBD, like the the radius part in this image.

G

The results are pretty the same, like sneaker.

G

But uh for 3v that I say uh I have tested this program three years ago, so I don't think it's a really a good, but at in the past the main problem of this type of scanner is that you use only the CPE and, for example, if you see this CV, for example, you you have a minor patch for Debian and you can and if you use only data for uh come from the nvd. You have just this part of.

G

So you have a lot of false positive with this approach and because the distribution backwards, some fix on the on our on the package from Debian, for example- are going to do the same. So with our approach. We don't have a lot of false positive, so it's pretty good for for our customers.

D

B

D

Running with the type, so thank.

B

You again Lauren um hippie to heavy um in any time if you need any help from the community um next. One on the list is give me a second: it's from Lucas yeah.

J

B

J

All right, so I will make a very quick, introduce myself first I'm Lucas at working at TNG technology technology consultant, um but I will not sell you anything today. Only one pull request, which is for free, I erase the pull request like a month ago, because I started implementing some Google application against the go. Harbor and I use, go client and I found it very useful.

J

I will share my screen real quick, um so it's maybe easier for you to understand what I'm talking about so I'm talking about this tool. It's very very, very simple.

J

It consists mainly of a make file where you can from the Swagger definition, create the go SDK and what I did is I bumped it to the latest version, and you see that there are many many files removed, and this is because the from what I've understood this Legacy and and the other older Swagger Jason Swagger yaml files- they don't exist anymore. So, instead of assist version, 2 and version, 2 Legacy there's only version two left, and this is more or less everything. I did in this pull request.

J

Right, I, don't know how to continue. Volume asked me to join the call here, so that's why I'm here I can only ask for a review and tell you that I really appreciate your work and, however itself, and it's always a pleasure working with it. Thanks.

B

Thank you. um Do you want to comment to this one as we.

C

I

Yeah so I invited Lucas, because I think it's uh would be a valuable discussion to see that we can provide a bit more love to the Google client library to go to the Google client, especially updating, because the the other background is that um for the CLI client that we or actually is currently working on, he has also discovered a problem. The go library or the go client and specifically, is I.

I

Think, as you mentioned it it's about return values so that if there is an error, the function doesn't return the arrow nor the error code. So it's just yeah, so he mentioned that.

I

There's also this the issue there and I think we could kind of a you know see how we can improve the Google client library, because we are also using it internally right, and so we need to find a way how we can improve it and I still be compatible on the client side. On the consumer side.

C

Hi modim I record that we have a program on the gold crime or and should do. We need to consolidate the either to make one group to focus on the Gold Crown implementation.

C

I

So I sorry I didn't understand it can.

C

You repeat: it uh I mean I record that we have a program. I mean uh um monthly uh mentee program that to implement the uh gold clan for Harbor.

C

Oh okay, okay, okay, so.

I

He's he's implementing go Cai CLI, but he's using the go client for that.

C

Okay, okay got it.

I

um So Harbor is using the Google Cloud as well right, it's not just exposing the call client to the our users. It's also because we're also using it right or.

C

The operator is using this: the.

I

Operator is using.

C

I

C

So so can you please correct me uh if I misunderstand something.

I

Yeah, so you see Lucas, we need to update uh if we update the go client here right. We need to make sure that the clients who are using the go client are also have the opportunity to update as well right. So we need to kind of uh not synchronize it, but you know they push it somehow Downstream and telling them to use the new go client Library so that they can update their code if they need to if they want to yeah. So how should we progress with the pr.

C

uh Sorry, I I missed this PR before.

A

C

Because so um yeah we can uh start to review this um vrs and yeah. So um it seems um reasonable for me to to ship another release to uh align with the Upstream.

I

J

All right thanks so today in the morning I updated to the today's latest version, so we should be fine. Now.

C

I

I'll check that yeah.

J

All right, so thanks for having me, uh if you need anything else, you can reach me in the pull request.

C

Yeah sure thank.

B

C

B

Thank you very much. Thank you for the effort um in last but not least, Stefan your head and your radius problem.

F

Yeah I'm asking as a harbor user I'm, not as a contributor or developer and I'm, not sure if I'm of if this is the right um stage, to ask this question. But I have opened an issue a couple of weeks ago and um just want to ask you if you can point me just in the right direction to go further with this issue.

F

um But if you you can stop me if there's not. If this is not the right stage to ask such specific questions to him.

B

It's fine: uh we try to combine the community meeting plus office hours and one so we can answer such questions as well. Okay,.

F

So um can sure my screen to.

B

Yeah better, if you share, you can navigate more yeah.

F

B

Through the issue so just give.

F

F

D

F

B

F

I can scroll and zoom in a little bit yeah. Yes,.

A

F

Is better so so yeah? It was um yeah quite some time ago and um we still using in Harbor with uh chart Museum, hopefully not not too long, but we are still using one with chart Museum and we have um one project within our Harbor instance which cannot be accessed. You can either pull images, push images you can delete it via a portal or API. You uh yeah, it's it's not usable, so to say, and uh we've gotten a screenshot here where we receive in 500 HTTP 500.

F

It's described in the issue more detail in more detail, but uh at the bottom of the of the um of the issue was that there was one suggestion to delete manual, some redis cache entries. uh So the suggestion was, if we looking for the key chart in our redis database and yeah, delete this entries.

F

Maybe it should work again, but um uh so so we are not really experienced with redis and how it interacts with horror, and since we cannot reproduce reproduce, reproduce this or replicate this in our test instance, we have to do this in production, so we are a little bit cautious. You know- and we are asking so if we are using if we are searching in our redis database, for the keyword chart we're receiving these results.

F

So um if we are good to go to just delete them and um without having big production problems or is it, is it safe to to delete the these entries, or um do we have to to be more cautious and find another way around it? Just a just a Direction.

F

Hope hope this makes sense what I said right now.

D

Any learning job I think you can think about all that. That is there and then restart their Harvesters.

F

Okay, so we are going to go to delete it and then restart our Harbor instance.

D

Yeah make sure there is no running job service series.

D

F

D

F

um It's just just for one project with having this problem in our harboring sense, where we have currently over 80 projects running- and this is only one which cannot be yeah accessed- deleted rebuilt um yeah.

F

um Maybe this is a good uh context as well, so this is not an internal redis deployed with Harbor hand. This is an external red database from our cloud provider.

F

So not not easy to restart it.

C

I'm, sorry, uh what's your uh Harbor version? uh Currently it's two six four, so this is for. um um Is it possible that you can just remove Keys from database three three um in release, because this is a specific for chart Museum, so you do not have to restart Harbor. You can just clean the db3 of radius.

F

Db3 is this another key which I can looking for database.

I

C

D

Yeah, that is multiple database.

F

Okay, so database free and redis, okay, okay,.

C

um I pasted a link into the chat. So could you open that so so they will see more details.

A

C

um Yes, this one, um you will see uh the there are four or five database for radius is live. One A4.

C

By 1 A4 one, eight four one.

F

C

F

C

Yeah yeah, you will see uh yeah, there's the TV indust and you can clean the all keys for database. Three, that's okay! Okay,.

D

At least it may be depends on your installation title for the top components you can use this uh reference. Other four channels in they are uh inspiration, type may be different.

F

Yeah, yes, we are using the harbor hem chart and then h a word setup.

F

Yeah, but this is a good direction, so thank you for that. uh We will follow that and uh we'll let you know in the issue if this was helpful and solved our problem.

C

F

C

So, do you have a plan to migrate the the charts file.

F

As I described in the last meeting um so um with the So, currently our teams um receiving their Helm charts, classic ham, charts uh Upstream uh via our implementation of artifact Hub in Harbor, right so they're, using their bitnami and kubernetes charts and so on, and um for this iteration, which ends on July for our team, we are planning to deploy. The oci Hem chart proxy from that in in our production system. So we are testing right now. It seems very good, um but I have this.

F

In production we have to yeah, be careful on testing in QA. We are planning this uh roll out in production next month and uh with that setting, we have the prerequisites to upgrade to 2.7 and then to 2.8 and deprecate chart and resume for all of our teams.

F

This is our goal yeah, but we cannot do this right now, because a lot of teams using classical hand charts so to say in their own projects and in a central project, and we need to support, chart Museum, I think for at least one or two months or three months yeah. Once we have a good alternative with the hand proxy, which seems quite promising, then we go, then we can upgrade to 2.8, which is our goal.

C

Yeah: okay, yes, um and we have a a tool to help you to migrate the existing source and paste in the chat, yeah.

F

Yes, yes, thank you. We will use it and we will give this to to in the hands of our internal teams, um but for the Upstream charts, because the teams uh should not download on their own from their own environments, uh charts from from every instance. So we have this one, a single proxy, which is our Hardware instance, and for that we need an alternative, and this is the hand proxy yeah yeah and it's running in our test instances and it's doing its job padding. It's quite good here.

F

Yes, so thank you.

B

Thank you hope that will fix your issues um within two minutes after the hour. uh So one last thing this Sunday. Yes, this Sunday is the end time for the cfps for kipcon, both kubecons China, so Shanghai and Chicago. So if you're planning on applying for a talk, don't forget the end date. If you need my help to review or give you ideas on your talk, your abstract, uh please do contact me, so we can work it out. Yes with him.

I

Okay, no no comment. I need to plan my weekend. We.

B

Can submitting uh we can not submitting cfps right, okay, yeah, okay, cool, so thank you very much. Everyone have a great rest of the day or evening and talk to you in two weeks and I hope. I won't forget what I've put it in the next agenda and so.

H

B

Thank you bye. Thank you.