 Hello, everyone. Welcome to the Jenkins Infra weekly meeting, infrastructure team meeting. We are today the 16th of January 2024. Right now we are only three with Kevin. Let me write you down, Kevin. Bruno, wait. No, it's not Verachten, it's Veraction. I forgot. Whatever you please. Bruno, yeah, that's good. Bruno. And for now it's all, all folks. We may be joined by Marc and Erwe later. As for the announcement, the weekly 2.441 went out today was fine. I think that we need someone to spawn the tag for the docker image. We'll have to check with Marc or Erwe. I'm not quite sure I have the rights to do that. But for now, we can see that the 2.441 is out and with its changelog and it's today. So it's fine. Announcement. Do we have any announcement, guys? Nope. Except that Damien will be off not next week, but the week after and we will still need to replace him for the voice for that meeting. So that's a value in an announcement. Upcoming calendar the next weekly will be 2.442 and will be the 23rd of January, like every week. Except that the next day and Wednesday that will be the LTS happening the 24 of January 2024. Do you have that number, Kevin? It may be based on 2.440. So the next LTS for the one on the 24th will be 2.426.3. Of course, my bad. Yeah, I think Bruno what you're referring to is Marc started the discussion about the next baseline selection. Which was supposed to have been started this past Wednesday. Yeah, the 440 is currently the discussion for next baseline. Perfect. So, let's see if there is any infrastructure security security, sorry, not infrastructure, but security advisory. Okay. Nope, nothing. So, no. Next major event. So all of us guy will be at the FOSDM in Brussels, the 23rd and 4th of February. The FOSDM by himself will be the three and fourth, but the two will be the contributors submit for Jenkins. So if we if you want to meet us just come to Brexit. And that's for the scale in March 15. I forgot in which city do you know which city is the event of the scale. It's in the US, but I forgot. Yeah, I can't remember the city. I think it's in California. Okay, so guys, if you want some information on that Google is your friend. Yeah. It's not a big deal guys. Let me start with the with the what we have done this week. Let's start with this one. Oh, that that the person that opened that issue closed it by himself. We asked for some some information to make sure that everything is fine but no news. So it's done. The user that had been blocked. Oh, that's a spam. Okay, blocked. The Jenkins CI to okay. Last permission. Oh, so Alex and not my fault is Alexander Brenda Brenda's. He took care of that. Thank you, Alex. You can meet him at the FOSDM to will be there. He wants to have the maintainer issue same Alex did and that's great. Thank you. Condition of Mark and that's okay. Oh, that one. I followed a little and Damian cleaned up everything. They just forgot to clean up the ACP to make sure that they change were good. And they tried again with the ACP flushed and it worked. So everything is fine. Someone have been aided. Yes. By an agitation someone have been eating and everything is fine. Good. Oh, this one is a big one. We had to handle that with with Damian. We found two problem with the the redirection for the mirrors. In fact, there were there's two problem. The main one was a little mistake from Damian. During the last issue where he updated the charts for mirror beads, the legacy one, and he forgot HTTPS. So the URL that was built was completely crazy. Like this archive is under mirrors. That's because of the HTTPS missing with that, that was working way better. And the other problem is that the asynchronous that is handling our failover system is is asynchronously updated, meaning that during that time. You can have four or four error because the failback is hitting the main one. So we have to change the way that fallback is provided. There is multiple sway, but we we need to make sure that we don't have that little gap where what is four four. It's not even 30 minutes that's depending on how it's handling the synchronization. So we will have to deal with that. Oh, and this one is mine. I did finish to separate. We know have three goes file one goes for Linux one goes for Windows and one goes that is common for both of them. And and with all the version tracked with the CLI. So that new way of working should provide us an easier way to manage the updates and and and the testing of everything. So it's done. No more girls for a while. Close as not planned. We got someone who forgot. Yes. Oh, it's local host. It's not us. So we cannot deal with account from another instances of the CI. And this one. Oh, this one. They did try a lot of things to explain why their their court test was failing. At the end they choose to increase the time how to 500, which is not really a solution, but to work around. Damian did look into the data to try to understand the the main origin. And there is no. Nothing pointing exactly. They were thinking of a version of the agent. They tried the old versions and had the same problem. And then with the new version, the problem disappeared. So it's pointing on a crazy network issue. Damian raised a new issue that we will see soon and it may explain everything. We'll talk about that later. So work in progress. Oh. So Alex asked us to remove one of his certificates. And have is taking care of that. Yes, we need to rebuild everything and we need to find that certificates. Good. Someone triggered it in this pan. So waiting for an answer to mark to deal with that. Same. Thank you Mark to take over all that. So that's that's a problem that we found a bug that sometimes the images are rebuilt and shouldn't be. And that's what I've been done. So. So everyone is on there. And Damian plan to removed. And clean. The bad images that have been rebuilt and shouldn't have been. So cleaning in progress. This one is forever. The aim is to split split sorry. The image of weekly, because we're using that Jenkins weekly for Jenkins Afra and weekly dot ci dot Jenkins dot IO. And we want to be able to deal with the plugins and and what we put in in each of those instance. In order to keep weekly as show you say show off. No, it's not show off proof of concept and and and showing the the latest improvement in UI and plugins so weekly is dedicated to that. Whereas infra dot ci dot Jenkins dot IO is the instance that we use to deal with infra and could have the same exact plugins and configuration so we need two different images to deal with so air base taking care of splitting that we'll see. Yeah, the idea here is to handle a version of of docs docs dot Jenkins that IO that are versioned and and to keep track of everything so it's every dealing with that too. Oh, this one is there's a problem on the database. So we need to repair to find exact the exact lines that are not correctly saved and to remove them for the database to work correctly and Damian did to start working on that. And try to find the correct line by the economy. So work in progress to repair the database. Okay. So in terms of memory. I don't remember exactly what happened on that. There's some failure. Dependent button the request. The problem here is to make sure that it's not related to any network issues. So that they are pointing on Java 21. So that may be not be related Java space. Okay. So I don't know who is dealing with that but it's a work in progress. Migration leftovers. I have to say it's it's most in my pocket that one. It means migrating data. So it's not something that you do in two minutes and we need to plan and to take care of that that was two pair of high and we cannot really spend time on that right now is Damian not really in so that we probably would have to wait a little. They are really important. So, working progress. Oh, this one. We did. I did the skeleton. So right now we can export a textual presentation of the URL used for the mirrors. And the aim is to to comply at and to take an example of this kind of data, meaning that we can have in one json file, all the data we want, for example, all the IPs of all the mirror can be in there. But we can also add others. Informations about the infrastructure for for us to have that infrastructure, a publicly available and up to date. So we need to convert my skeleton to json and to add information and for example right now information are URL we want to have IP so we need to dig on all the URLs and add more. For example, how IP from the infrastructure coming from the terraform can be pushed in that json file. They took over. So we should update that, but are they took over on that json file. This is my current work. For this, and in order to use RM 64 more and more. I am moving all the image that we're using right now that are not RM 64 to the all in one in RM 64. And so I have to improve the image with all the tools that are provided by others images. And right now I'm dealing with the one used for the terraform. And putting all the all the tools in the all in one, and then trying the new version of the shut pipeline library with each of those repository and to make sure that they are going through and working. And we'll have to do exactly the same with update CLI using another image and to make sure that the docker N5 and to make sure that that image is correctly replaced by the all in one. And with the all in one we will be able to work in RM 64 or I am the 64 pods or VM everything will be exactly the same because they're built all in one. And that's the aim of having only one to update and keep up to date. So right now I'm working on those, the shut pipeline libraries. One, it's a VN Damien, they try to replace blublix for us with as you copy, because that will solve a lot of mislinked that make think that we don't we don't download the correct file because it's not synchronized correctly or it's synchronized too much files and then changing their dates. And I think it's costing more money because there is writing on the side. So no, no progress yet. They're working on the service principle, because we don't want to to give too much trust in those contents in those. Sorry, where we we copy the data, if you if we give too much rights, anyone can hack and put stuff in those. So we need to make sure that the security is really closed and and no one can access that and put stuff in those. Okay, so this one is postponed and is waiting for security team to have a look on the code. If I remember correctly, we are at that point. No, no date this week. Now we still need to make sure that service principle is is dealt with and same with blublix for same. But no, no progress and still waiting for the security team to make sure that we are not doing anything crazy. And Damien found an issue yesterday on Azure where we got a SNAT port exhaustion network. That's something nice. Explain the networks problems because for every connection to the outside to the Internet, there is a port opening and that port is usually open for 30 minutes or one hour, depending on the configuration. And we only have something like 64,000 SNAT and we are eating that limitation. I think it's pending total. It's per hour or something like that. And this is and this is causing network latency and network slowing. So Damien think that he may explain those issues. And we need to. There is multiple solution. We can move CIG and Kinsai on the new subscription and it will have its own IP for the outgoing connection. So limiting the volume. We can also change the configuration and have a window of those liveness of those ports shorter like few minutes. We can also and that's what I think they have done with the old ProdPublicay 8S having multiple IPs for the outgoing connection and using a lot balancer to use on the outbound connection. So there's multiple way. But we will have to deal with that and that may solve a lot of our problems. But migrating to the new subscription is something kind of easy and that can allow us to use more of that subscription and maybe the first step. But we'll see. And yes, that problem on Windows agents much slower. We don't have much information, but this may be. Involved to is the network connection. When it's it's slowness week. It's not easy to pinpoint the problem. So. When when the network will be network problem will be solved, we will be we will then be able to pinpoint which problem can cause slowness anymore again still still. Yes. So we are done. We are done. Let's see if there is some new issues. Let me three age. This one we spoke about that we spoke about that. This one quote the end for next execution plugins. I'm sorry I don't get this issue at all. You. It's about adding this plugin in the crowd. The first administration. It's an ASS to manage translation. Okay. Hello. Let me know you did that already. Yes. Thank you. No problem. So do you have something to add because. We're done. Yeah, thank you. So it's all for today. Oh, I can I ask you have a we I checked the two four four one of today and it's out. Oh, Mark. Mark. You created the entity tag about 20 minutes ago. We should see the container image in. In 20 minutes max. Perfect. Thank you. So tag tag will be available. So no docker image. Will be available. Shortly. Thank you. So if nobody has something to add, I will tell you. See you next week. Have fun. Bye bye. Bye.