 Hello, everyone. Welcome to the Evergreen Open Plan meeting. So weekly meeting that we have to every week at that hour roughly. Even when there's no time zone changes. So this week, I'm going to handle the thing. I'm going to just start sharing my screen right away on that. Oh, I can't do that. You haven't made a share screen? Oh, no, you got it. Yeah, I was trying to share a single screen, but for whatever reason with I3 wasn't working on that one. Anyway, that's a big issue. So I guess let's get started with you, Mandy. How was the week and how? Well, I've only been minimally involved right now. So I am working on fixing the bad tests that you pointed out to me. Good job. By the way, where is, why did that happen? Appear here? To take off the exclude clothes, it's one of those weird gear things. Oh, because of the fact it was fixed and then back to in progress, but it's taking in account probably the fixed thing. I have to get rid of that state. It's very frustrating. I think you can, but you have to reopen probably instead of just getting it back. Where is it? It doesn't matter. I'm working on fixing that and I will have it fixed and checked back in today. So that's that basically for people watching it. So that was not the one, by the way, it was more that one. So it just showed up. Okay, because it's closed here. Okay, it's really weird. Anyway, right. So great. Thank you. Anything else? Not right now. Thank you. The make check fails on macOS. Take it that you've got Mandy. Why don't we just unassign that and maybe we can convince somebody, somebody else in the Jenkins project to help with that? Because I know Batiste and I don't have max. Yeah, I'll go ahead and unassign the ones that I'm not going to get to you anytime soon. And please somebody else feel super motivated to pick them up. I'm sure there are plenty of people who want to fight with doctor on macOS. Really? No, it's going to require a significant resectoring of how we're doing doctor because everything from Mac has to be explicitly spelled out. And the last time I went down that road, it broke my head. I wonder how long and hard it would be to fix it now because I have never tried it, but I remember in the very beginning, so it was less convoluted in a way, had less tests and setups and everything. I remember Nicola Delov fixed the make check that wasn't working on macOS for some time, but I guess it's failed. It's now failing again for some other reason. So yeah, so yeah, last week and I spent the same thing and same time last week actually unassigning myself too. So now I think we have a bunch of things that are supposedly planned for my stone too, but that aren't assigned. So anyone watching, feel free to reach out. So for context, we have also tagged many of those as newbie friendly, which is the things you probably want or can dig into if you're interested into a hacktoberfest. But if it's not tagged this way, I mean, someone can still reach out to us and get guidance that things can be tackled because, well, we are pretty much very, very eager to have people. So we'll be welcoming anyone who want to try to give it a go. That's it. Tyler, your turn. Yeah, I'm super behind. Most of my evergreen time has actually been consumed by Daniel as uplink time, fortunately. So I know that Olivier was going to take a look at the read on my database access for y'all. So I think that's, yeah, he's got that in there. I don't know where he is on that. In the info status meeting yesterday, we didn't discuss it because it just completely slipped my mind. Yeah, from what I know, because we've discussed that way, the idea is going to be so there's no possibility to open easily Azure to some set of IPs or something like this. So what he's doing right now is to set up a gateway machine so that we would then make runs in that machine, a Docker image, with pre-provisioned public SSH keys so that people are allowed to go and to reach out to that DB. Their public key would be there and then be allowed to connect to that gateway machine having access to the DB in read only mode. So we really want to make a check here, Mandy, because I provided to Olivier a link to my own pub key. But if you want to do the same, I guess it would possibly help save an iteration there if you get access directly. Okay. So he's got that going on. The only other thing that's say a little bit block-like right now is there's these telemetry analytics and pipeline use of telemetry things. I'm still waiting for the pipeline developers to give me a sense of what they actually want or what they would actually need before moving forward on that. So I'm going to add a flag because these are all kind of blocked. So I guess you mean the 49852, that one? Yeah, there's three of them. And now they all have flags on them. So if you refresh your board, you'll see that. Oh my God, flags. Flags! But in parallel to that, I've been... This is related to this work but also to the uplink work that I've been doing with Daniel. Been researching some other open source projects privacy policies with how they handle the sort of telemetry data. I'm going to go ahead and put the KDE one in the Gitter channel. This is likely what we'll build off of. I think KDE actually has done a pretty good job on this. But this is probably going to be where I'm going to be spending my evergreen time this week is actually forming this into a job and explaining it to people. Because as it stands right now, technically I can grant access to some of the telemetry that we've been collecting and we've got this information going into Century as well as you both know. But we don't have any policies for how that data should be used or expectations set for users on how that data is going to be used. And we need that for both contributors and per-end users before we start putting a lot more, I would say, rich metrics and telemetry into any of our project systems. So did you, for instance, understand a bit more about, for instance, GDPR and the fact that will we be forced to have some kind of when the image is started, you know, some kind of agreement or we'd be in the wizard? For GDPR compliance, I haven't had a conversation with someone specifically about evergreen on that. I think what we are probably going to want to do, regardless of GDPR, is once we have a data retention policy, make that known to the end user. I don't know if we can come up with anything. Personally, I think these are stupid, but I don't think we can necessarily come up with a click OK to acknowledge and to agree to this sort of stuff. I think we have to make that more clear before a user starts evergreen. And once evergreen is up and running, what sort of data is being collected? But I don't know if we can have an explicit OK modal checkbox. The biggest concerns that I have are less about evergreen and more about uplink telemetry and our data retention policy is there. So evergreen, I think, is actually going to be a little bit easier. Uplink is less easy. OK. Policy, that's what I'm hacking this week. It's so much fun. Managerry things? Yes. Anything else? After last week, I know we said, our last meeting, we said, hey, Joe, if you're watching, send us an email. I haven't received an email from Joe. So we don't have any updates on an ECS-based cloud formation template. So there's that. Finally, I understood what you were saying. So you said you received or you didn't receive an email? I didn't. OK, I didn't. OK, I did not. I thought you said I did. OK. Right, right. So we'll probably just do a grip or something to fight all Joe's around San Francisco. I don't know where he was located. That's all just explained to the US. He was an American. So there's some Joe in the United States. That is the Joe we are looking for. Yeah. Indeed, you know some Joe's? Anyway, so I guess that's all or I guess you could talk. Is there anything else you want to talk about, Jenkins in general? I don't know. Not for me. You could. Not today. So I'm going to just hide closed ones where I am here. So the status right now I've got a bit more to talk about than last week, which is dangerous for people watching. So the Evergreen Automated rollback is now in review and almost review that would say about to be mergeable or merged, which is that one on the right. Then there's so this I've segregated two changes here. So that one is really about updating to a new update level and then realizing that Jenkins wouldn't start after it and then actually automatically running back to the previous one, but that doesn't handle any data format change. So if, for instance, that's what I'm now working on, there's something going wrong after an upgrade and something that that format has been upgraded. Then we go back and everything is screwed up because it's not backwards compatible. That's the other one I'm working on in the very center. So the 57 at the end. But that one is blocked right now kind of, but I'm still moving forward globally on that global story, which is that one basically. What is it blocked on? On that one. Oh, they're just dependent. Yeah. There's a blocked on dependency between that one and that one because basically I'm right now working on something that will make me able to test that my rollback is actually working because then if I don't, then I could just rollback and be happy, but it could just not be working. So I've been spending and getting a bit crazy on that for almost the whole day because believe it or not, I've not been able to crush Jenkins. We were talking about that in the chat just before. I'm maybe going to try another alternative way. Even if I do know some ways, but for some reason I wasn't able to do that, but for instance, working on purpose, the config has got files between... If you need help crashing Jenkins, I know Carl Schultz. I think he might have some ideas for you. Yeah. So my main purpose here is to exercise the one below the 54.057. So basically I want to make sure the snapshot, the data snapshotting system, and so using it as a restore system works. So this is somewhat specific. So what I was trying the whole day was to have some plugin that would change its data format and upgrade from one version to another and then back and have it crash. I think I'm missing something obvious, but the good thing is that I pinged actually people knowing about Jenkins a bit earlier in the day, and JC told me that it wouldn't be easy, so I wasn't getting totally crazy and I would have to go out. Even JC said, well, it's going to be a bit hard. The alternative for doing that actually might be simpler if I go another way, which would be actually by using Docker exec during our tests, breaking or changing some config.js files, because we know also a config.js code is written in a way that it actually crush Jenkins during startup if YAML files are incorrect. So that might be a way, maybe a better way. I'm not sure. Anyway, overall what we want to test is, indeed, if we are to actually be able to, you know, go from one level to the other. For your purposes, what do you need to see from Jenkins to indicate a crash of Jenkins? Let's see it the other way around. I need to not see it. The JEP 306, which describes how hash checking is done, basically, I need the hash checking to be wrong. Okay. If it's not available, does that count as wrong? Yep, because hash checking then will fail after a timeout. So we'll consider that Jenkins is unreachable and then revert all back. So one thing that you might consider trying is with Jenkins, if you have two Jenkins instances running with the same Jenkins home, the Jenkins will fail. And it'll fail to boot. Like, it doesn't even boot properly. So there's some file, and I can't remember which file it is that indicates to Jenkins that it is currently running. It's like a PID file or something like that. For your test, it might be viable to just artificially create that file so that Jenkins thinks that another instance is using that directory. Yeah, that's a good point, though. It might also reveal something we don't cover, because actually I had a look at that code recently, and I don't remember exactly which classes, but I think it was someone asking for it on IRC or something. But basically maybe that file is actually out of the Jenkins home, maybe on top, probably there, yeah. Because I was thinking, because the JEP 302 is only, you know, snapshotting the Jenkins home. So if it's like under the Jenkins var, then things wouldn't be reverted. But yes, that's another alternative idea to actually touch that file. Anyway, I think I'm going to try a few alternatives and should be finding something at some point before the end of the week. By the way, about that, that makes me think on Friday of this week I will be in Copenhagen just for the day to have a talk about... It's the day of Jenkins, isn't it? Yeah. Woohoo! Yeah. What are you presenting? Something about Evergreen. What do you think? Are you not sure yet what you're going to present? No, no, I'm kidding. So it's the day of Jenkins. So that's that thing. And so I am presenting basically Evergreen. I know some of these faces. Yeah, some. And so I'll see how I do things until the end of the week, but I'm not sure I'm going to try and timebox the time I spent on the snapshotting, I think, because I would like possibly to table that for some time and maybe try to get AWS Flavor back to working, because it might be interesting to show that during the demo, during that conference, and overall useful for users and everything. So yeah, if anyone has any opinion about that, that could adjust my take, but yeah. And after the... Well, I'm thinking about that. After the conference, I'm on Hoide for one week. Anyone here? Yeah, right here. Cool. So we are on time or we are running out of time basically. So anything else before we close? No, I don't think so. Not the moment. Thanks. So thanks everybody. And so I'm going to stop the broadcast and bye-bye.