 Well, I don't know that we have a huge agenda. So should we just get, All right, let's see. So we got on the agenda bots. Yay. I'm actually really liking this whole stale bot thing so far. It's cool. Yeah. It's on the, it's mostly bots and bots, all right. So I don't know. Should we just start, start with bots? Oh, Christopher, you want to give us a little update there? Yeah, I didn't have much to say that I didn't have a fair win there for just a safe time. So yeah, glad to hear the stale bot's working and thanks for merging that PR. So now everything should correctly get marked as stale instead of a, you know, old fix. Yeah. I actually went in and I untagged all the ones that it had tagged as won't fix and I marked them stale, but in actually marking them stale, that counts as an event for not being stale. So it'll take another 30 days, but that's fine. Oh, okay. The other bot that we wanted to do didn't seem like it had a clear hosted one or if it was a hosted one, it was, it didn't include all the things that we needed. So I have a work in progress up for that Envoy Bot repository, which allow you to do like a forward slash a sign and then give a couple of GitHub names and then it'll also verify for members and comments if they are not back to the issue. The weird bug that I found is that running into some weird JavaScript, like async away problems, it's whenever the GitHub user just does not exist at all and it keeps commenting back and just throwing it into a loop. So that's kind of the last bug I need to fix before I'm ready for the PR to be fully reviewed. Cool. Sounds great. If you're super ambitious, there's a lot of changes that I would love to have on the DCO bot. That would be awesome. Yeah, I did look into that part of the issue and I kind of agree with Alistaz's comments that it seems like one of those like specific legal lease things that makes it well, there's just like to modify it unless they are explicitly opting into. Well, sorry. So I didn't actually even mean to do the auto DCO. I just meant to fix the bot so that it like actually has error messages. There's a bunch of cases where it just hangs and doesn't print any messages. And there's some basic stuff that we could do there where instead of just saying, you failed, it could actually link you to a webpage which actually explain what to do. So I think that there's just some basic usability things there that if we did would be pretty awesome. Okay, yes. I think we open up any bot related issues on the Envoy bots repository. Sorry, what's that? Should we open issues for the Envoy bot or any other in the box? We can do that though. Though I suspect that we could just submit PRs directly to ProBot DCO. And they would probably review them. Oh yeah, that's a good point. All right, yeah. I will look into that as soon as I fix Envoy bot. Thank you, that would be really awesome. Is it straightforward to have a button to just, we execute CI in cases where it failed due to not being able to download some dependency from the network? It should be straightforward. It probably just depends on, we might have to invoke some Circle CI API call, but I think it should probably be pretty easy. But yes, that would definitely be awesome. And I'm not sure, at one point, somebody maybe suggested that to take that off, I could close and open the PR, but it seems like it optimizes around that sometimes. Yeah, it does. I think that doesn't work because it just figures out that it's already built that particular shod. The thing that will always work is you can push an empty commit and that will always kick it. The downside of that is that that will rerun all tests, which can take a long time. So, I mean, like most systems, like internally we have bots that do all kinds of things and one of them is just a re-kick test that failed. So like, I don't think that would be hard to implement. It's just someone would have to look at the Circle API and probably figure out which API to call. But what we should do is just go into the open bots issue and just write down any and all ideas and then hopefully we can find someone to eventually work on it. If you just search issues for bots, there's a bots tracking issue. Very nice that we're making some progress on this. Anything else on the bots network service extension update? Dave? So, I'll start with an announcement that I'm in the process of moving to a new opportunity. So, my last day, it's just go is next Tuesday. And so, I'm in the process of doing a bunch of TLI work and so, the work that I've done so far, I'm handing off to Ed Warnacky and my relative is gonna get recruited to work with him. So, that's kind of why things stalled out last week and I didn't get the stuff pushed up that I was going to. In any case, there was, I did see and mention some place of somebody wanting to remove all of the native socket calls and use the OSS Singleton to do that. And so, I was actually thinking that was a great idea as a place to start because I was heading in that direction anyways. Yeah, that's fine. I mean, I don't think anything has changed. It's like someone just needs to go in and just make all the changes basically. Yeah, go do cleanup stuff. So, yeah. Cool. Well, good luck at your next gig. Thanks, man. All right. Small announcement from me. I am gonna be working on a non-Envoy related fire drill at Lyft, probably for the next four to six weeks. So, I'll still be around to do code reviews and stuff, but you can expect me probably to be a little less engaged than normal. What happens? What's that? Yeah, well, Corbin and I are also in the same area. Yeah. The upside, I can't talk about it right now, but it's actually a pretty entertaining fire drill. So, I did actually volunteer for this, but yes. There's that at least. All right. Anyone else have anything that anyone wants to chat about? I'll just mention that we are looking at downstream circuit breaking. Awesome. We care about a lot and are gonna hopefully solve a non-Envoy. Yeah, I mean, I think everyone cares about it. No one's just implemented it yet. So, it would be really fantastic to get the implemented. I talked about it with Alyssa offline. I think it's pretty straightforward, but if there's any questions and we just wanna agree on the overall design, I would just have whoever's working on it just update that GitHub issue. Yeah. Yeah, we will. Alyssa added also timeout switch. I saw you kind of on that board. Yeah, that one is, that would also be great to do. That one is trivial. So, that would be a good starter issue. So, if there's someone that kind of wants a little ramp up issue, that would be great. Yeah. No, I believe we need to do that in the order of the next couple of weeks. So, we'll give it to whoever can do it soon. Okay, awesome. Yeah, and that will include the data frame timeouts also, right? So, we've got two open issues tracking, one for the year making for our best timeouts and the other for the timing out and their request paths. Yeah. One of them were okay. So, we'll probably do one of them quickly and then leave the other one for a new year. Okay, that's fine. Yeah, I mean, I think probably then the best thing for you to do is just to do what we talked about, which is deprecate the timeout in the buffer filter and move it into connection manager. But we will end up implementing at lift probably in the next like order weeks or one or two months the data frameouts because we're starting to deploy streaming APIs and we basically have to have them. So, they'll get implemented one way or the other. Got it, cool. Nice, great. I'm already, I guess since we have some time just to quickly chat out. How do we feel the load balancing of PRs is going like would the bots be able to sort of do it? Would that play a role? That would be really nice. I don't think it would be hard to implement. I still feel like it's ad hoc in terms of how much work each individual maintainer is doing. Like it hasn't been a huge deal just because I feel like the work is getting done. But it would be nice, I totally agree to maybe automate the actual load balancing. So like one thing that we could look into and this is something that maybe Christopher could actually implement once we have something that's hosted is, yeah, like if we had just some basic data store, like whether it be Dynamo or something else, we could just keep track of like the maintainer rotation and then literally just round robin, one senior maintainer and one non-senior maintainer for every PR that gets opened. Yeah, and I mean the initial assignments might not always make sense and we always exchange them. Yep, yeah. That would be great. I think that sounds fantastic. Does that sound like something, Christopher, that you might be interested in implementing? Yeah, I did notice someone implemented something like that. I could try to implement that node module. I just wanted to get code owners as well so that again for things that are extensions and actually have code owners, it goes in. Yeah, yeah, that would be really nice too because then if we did our own assignments, like I think the GitHub code owners doesn't support auto assignment if you're not a maintainer or something like that, but this would allow us to automatically assign things to code owners that aren't maintainers for extension. And also we could put in hints for the bot or things that we are more familiar with. Yes, exactly. And we could use the code based on the actual node. Yep, yeah, that makes tons of sense. So even for existing maintainers, if we know that we want particular people to review certain sections of the code, we can definitely add people for that. So yeah, I'm a huge plus one on that. That would be super awesome. You brought up the hosting. Maybe Chris and Asher could comment on where the CNCF have a place that we could host this like node bot. Yeah, we have servers that packet or cloud credits, whatever works, whatever it feels easiest for you. Yeah, I think last time we talked about this, we just decided to like pick a cloud provider and I suspect for our needs, probably a free account would most likely be fine, but we can set up by account and then just have CNCF pay for it. Sounds good. Yeah, so probably- Is that documented in like the CNCF doc somewhere? You would just file, we would just have the project do a service desk in Korea and then we would approve it. Yeah, so probably what would be easiest is just to open a new cloud account, whether it be like we can pick, it doesn't matter to me, whether it be Amazon or Google or something else and then make like an on-boy account with on-boy credentials and then we can get you access to the on-boy last pass for the credentials and then we can give Chris and CNCF access to the account and then they could just take over billing. Okay, yeah, that sounds good. I'll try it out locally, make sure that all works and then I'll file the service desk issue. Great, awesome. That's super exciting. Yeah, once we can, the only thing that I think is important is, and like we had talked about this a while ago, is whatever we do, we should just try to make it as seamless as possible. So for example, we should be able to make it so that we have a project, like we can hook up CircleCI to like run some basic tasks and then auto deploy and then we can host it in like a Lambda or something like that. And that would mean that it's pretty much just, people can just make changes and they'll just automatically get deployed. Okay, yeah, I'll wire that through too. Awesome. And I think people's preference is to use Python unless there's a good reason not to. So let's try to do that. Yeah, on that note, so ProBot is written in, well, it was written in Node and I think they just refactored it to use TypeScript. Okay. So everything I have at the moment is in Node, but I don't know if there's extensibility where you can script it to call on other languages. That's fine. I mean, like if there's existing code that we're using, I think it's fine, but like if there's fresh code that's written, I think the preference is to use Python. Okay, yeah, I personally I prefer to try to use the Python too. So I'll look into that as well. Awesome. Okay, great. This is great. Cool. Anyone have anything else? All right. Take it easy everyone. Bye.