 Hi everybody and welcome for this new Jenkins infrastructure meeting. Today, to the agenda, what we have, we have a small topic about reporter Jenkins here at Oregon. We have a topic about update center issues we faced last week. Some updates on the Docker image build that Damian is working on at the moment and a quick update on the puppet puppet code that I'm currently updating. So we don't have a lot of the big which is great and it means that it won't take too long. The first one, the reporter Jenkins here at Oregon. So we still we are still affected by a few issues. So, as a reminder, we noticed that since the 24th of June, different didn't upgrade of the service. And we were affected in the way we were we are using it so we were not able to copy artifact from between maven repositories that affected us when we did the security updates two weeks ago now. So we had to put in place a temporary maven repository as we could not update the official release maven repository the prediction one. So we put in place a virtual maven repository where we would be able to fetch packages either from the security one and from the prediction one. We are still affected by that issue so I opened a support to get on chief rock yesterday. I was in contact with them today and now I'm waiting for feedback from from them. So that's still an ongoing ongoing issue so until then until we have until we better identify what's the problem here. We still have to use that fold back maven repository. So that has several consequences. So every tools that fetch Jenkins or war from the prediction maven repository does not work at the moment and so we have to override with a temporary one. And so I don't know if I can quickly show you. Yeah. If you have issues using the prediction maven repository. Feel free to ask and I'll give you the temporary URL. So that's it. Any question. No, sounds good. So I can continue. So last week we had an issue with the update center. So because so we we really cash artifacts and what happened from time to time is if we try to download a plugin. We have a return of fall for a horse. We cash that that horse. And so what what what happened last week is we try to fetch metadata from plugins that were officially there. And that's because we had some cash containing the for for answer, actually Spencer, the update center job was failing. And because we are using a temporary maven repository. So the the fall back that I just explained. We are identifying the wrong of this center of configuration so Daniel came back I was available and beginning of the week yesterday. In fact, quickly identified issues solve that. What was important for us is we forgot the maven repository maven repository issues until we have a fix for that we have to keep in mind that we are not using the production one. So the topic is about building is building the current images for multiple architectures. That's something that they mean is working on at the moment so maybe they mean you want to share more with us. So, let me take back. So, there are two topics here. The first one is a pull request that has been just switched by mark weights to ready for review. That's a pull request that team started and I work with him collaboratively. The goal of that request initially was to make the build faster. And what it on the line is that we were building and testing all the Linux images on the same machine and we have the same issue windows. That's why most of our builds are really really slow because only one Docker engine. Sometimes it's on a big machine on a high memory. So it works very well sometimes it's on a smaller one. So not not only it's prone to instability, but it's a big constraint so the pull request not only parallelize the test harnesses so you can run one test at one time on the one CPU core. Even two per CPU core because most of network bound. So it's got it's going faster, but also we changed the pipeline. So now you we spawn one machine for one Linux platform. So you have one agent that build Alpine Linux and test Alpine Linux, another agent in parallel that builds and to us and test and to us, etc. And those is not part of that. So why mentioning this year, it's because a lot of the pull request on the official Docker image are suffering from instabilities during the build. Most of the time what is happening is that all of these images built at the same time until the new pull request is much are triggering out of memory error on the kernel of the agents. So you see on the big log console events. And this happened on the cases where you were you are not using I memory machine, you have no and most of the time, unless you split the platforms. And secondly on the high memory machines. Sometimes if the machine has been used for builds before on CI, you can you start seeing GNL P agent disconnecting because the CPU is completely use that 100%. So first work with it. So I hope that's parallelizing on multiple agent and the test should be good in terms of timing. The slower test is now test harness is now taking two minutes and a half, instead of five to 30 minutes. So that's twice to five time faster, which is really good. Windows is the next step. So the next topic, I'm going to start working on this one. It will be building the multi arc images to be able to provide build and publication of IRM PPC and as IBM said the CPU architectures. So I'm still reading what team did on the previous days. So I'm still landing cover. So it needs to be able to run safely this workload. Morris coming on that we should we will we are, we have an issue on the GitHub repository with the details. Just just one question. So do you mean that the admin to arch build which is so that PR is in draft mode at the moment is the one that is almost ready. We have already found multiple requests. Some have already been merged. One is we started we are we are already using Docker bake. That's a way to specify the collection of images on the file in HCL language. And we use Docker buildings bake command that allows to define things that were only available on the pipeline until right now. Even the multi architecture in parallel at the same time but we're still we are still searching the way to do it. We can use chemo for building at least that one is quite easy and should work. That means you only require a machine with Intel, but we have IRM machines. And we have the issue of machines between CI entrusted. In that case, we have different security issue related to the sharing these machines because these are, let's say, we don't have a lot we are one or two max. So the pull request multi publish multi arc from team in draft is about that system being able to build and publish both. Your question. Yes, thanks. Yeah, thanks. Last topic, which is on my side, so I'm currently doing some of its updates. So, I started one PR containing many different small dates it's kind of, I mean, we haven't updates the perfect code for why and so it's like each time you upgrade something it has consequences on the third module. So the PR is here. It was working. So, yeah, that's if you if you want to do some perfect code and you are available. Be free to join the party. So the initial reason why I started working on that was to redeploy archived Jenkins that I go, but it's like, I need to date that it's a quick module and because I have created it's a quick module, I need to upgrade other module and so on. So, that's, that's what I'm currently working and I hope to have something finished for the coming days. I will go listed to move and to close a rack space account. I think you have any question or do you want to bring anything that's a big I think we cover all the topic that we that we initially had just one I want to add we about Aditya we need to think Aditya you want high to remove in production the nexus on the relief name space. Right. So I'll switch the discussion and I see you're in private with you to think when we want to plan this and we will, I will publish a mention be in IRC before when we will have selected the time slot. Okay. On my side. If you have anything to add to that. I actually wanted to ask one question so I am facing this error so I just wanted to know whether this is due to the artifact out there. Is it because of that. So, which, which are you get. Yeah, you put it in the chat. Opening it here. So you can see in the book, I hope that. So, authorization failed. Now that means that you don't have the right username and password to push. Okay, so this one, this one, yeah, this one is a four, four, three, four billion so I can help you. So yeah, it just means that you don't, don't have so the GitHub action doesn't have the right username and password to push to map and repository. We can debug that together. I could get it made that full request, I think, and he's the owner of this repository, I think so we can put some secrets on the tower something like that. Yes, I just have because I'm not sure that I have permission on that one to do a high yes it sounds like I do have some permission. Yep, I should be able to help maven token. Oh, there is a maven token updated five minutes ago. Have a username. So yeah, it's definitely an issue with the credentials and another option maybe is. So, is it your credential. No, no, I don't have that. repository. We want to have when we need to grant access to someone to push to maven so repository. What's the name of the permission of data. So we just have to be sure that the username who pushed the artifact as a right permission so we just have to put to list him in the right so oops with conventional commits. So that was updated two months ago. And who can push here. Okay, so. Okay, sounds like I can push to the get to put to the maven. Okay, we just have to double check who me which credential was used here. But you can we can we can do that together. I have one last update that I would like to bring is I won't be available next week so I would like to just console next week infrastructure meeting. Okay, unless you want me to drive it. If it depends. Oh, or you could drive it. If you are available, let's then let's if you are if you are available then yeah, let's, you can drive the meeting that's fine, then perfect. So you have we have someone to drive the next week meeting how some so don't need to don't don't have to console it. Then I think we cover all the points so thanks for your time. Enjoy the rest of the day. And see you on RC. Goodbye. All right, thank you.