Augur/8-Knot Meeting Feb 17 2026 ![rw-book-cover](https://img.youtube.com/vi/WC5V6K_HTN4/maxresdefault.jpg) 🔗 **Source:** [Augur/8-Knot Meeting Feb 17 2026](https://www.youtube.com/watch?v=WC5V6K_HTN4) All right, we are recording. Wait. Um, I had a random random sort of first item on the agenda, too, was um we're under 60 poll requests now. So, that's good. I think between between Sean and I, I think we closed like 15 PRs yesterday or something like that. >> Yeah. >> Um, so that's that's probably the most most single day closed PRs for at least the last few months. Um, it's probably going to get harder because a lot of the other ones I keep kicking the can down the road like, "Oh, we'll deal with this later." And there's less later. But um yeah, I think uh like we like I was just starting to mention before we started recording, I think we um the main issue we're seeing at least from mine and Cali's side is um contributor emails are in the in the in the contributors table is a lot lower rates of them actually being sort of resolved and filled in than than we think they should be. Um I know Sean, I think you said you were running some tests to kind of confirm some of this aspects of our contributor resolution. Have is there any results from that so far? >> I'm looking at them right now. It's It's still running. Um >> was that the large test or the small one? >> The small one is what I'm trying to look at for approximate value. I'll share this real quick. So, this is a small database. Um it's got a by small I mean like 122 repos. Um we had 635,000 commit rows that didn't have the platform ID. But to your point about emails um before I started running it I and this is just looking this is just using the commit API to pull emails in. And there's two other processes that I'll run, but this one has we think is the most gets the most items or the most people taken care of. Um, so we started with 2516 that did not have an email before that ran and 649 that did. Currently, there are 1,828 that don't. So, it's not done yet. So, still 1828 that don't and up to 1565 that do. So, I think it's I think this is functioning at a higher rate than the approximately 15% that calling the the contributor API did for both emails and um GitHub ids. Now the the one thing that that is occurring that is um we are getting a lot more of the um platform usernames populated in the commits table. So that that part's working much better. Uh so we didn't have that as complete before. >> Yes, Kelly. Um, where in the process would it be that like I'm glad to hear that there's more contributor resolution going back to the commit site because I think that's important, but say like now we're starting to look at different avenues to get that email information and I think a lot of times there might be a like the contributor res resolution comes from email and profile, email and commit. Um, and so is there a way like is it checking it in that direction as well or is it only going one way? >> What it's So I think you're asking Let me answer the question I think you asked and you can tell me if I'm wrong. >> Why don't you say ask the question back to me and I can >> ask question. Okay. So I think I think you're asking if we're getting the emails from the commit record if we don't get a return back from the API. No, that is not my question. My question is is that if the for a given contributor ID, say we get an email from a place that's not commits, are we checking it in that direction? Say for example there is an email in their GitHub profile and we can tie that contributor to those commits that way from like matching it like that for some reason contributor resolution didn't work from commit to user but we have it we have that email from the user profiler somewhere else and does it check it in that direction? I think the answer is no. It's not doing that. So just >> and it makes sense that it wouldn't because we haven't like we're just putting in that functionality. >> Well, and the the commits like basically always have an email. What I'm learning from just doing this is in the last several years, it is more often one of these really weird not real GitHub emails that get assigned by the platform like it's like commit hash like your use commit hash truncate or appended to your username at github.com. There's there's a lot of those right now. What we are what I am doing in this test is what I think augur should be doing and is doing but I my theory is it's not getting triggered or that that's failing somewhere that we're not seeing. What I'm what this process is doing is saying goes through the commits table. If there's no login, it tries to get the login based on the email. If it finds the login based on the email, then it tries to match it to a record in the contributors table that doesn't have an email and put it there. If it finds the record has an email, it will check to see if it's the same email that it found in the commit record. If it's not, it checks to see if that email is in the aliases table. If it's not, it puts it there. That's >> okay. >> Uh that's what it's that's what's happening right now. And I just checked the progress. It's about on the small repo. about twothirds of the way through. Um, so those those results um I don't know. I think they're kind of promising because we're getting almost all the platform usernames assigned. So, if we can get the platform usernames assigned, we should be able to like I'm not sure if I'm I'm not sure if this process is going through the commits rec commits table um and checking to see if there's a a login and then going back to the I mean, it should be doing this, but I don't know the rate of it's filling in the contributor login on commits at a significantly higher I mean, there's more of them, but it it seems to be doing more of at than getting the emails back into >> the most of the time like cuz that means like we can that's a that bridge is much easier than the no longbow side bridge. um like we can like that's that >> we can make our I think this is a lot of I think we're realizing that there's a lot of like small problems happening and so if we can just kind of make solid progress like then you have it on the login like yeah that >> that's a pretty simple process to to make that last bridge >> yeah I think this largely tracks with what I've kind of seen like from our testing in the last week like it seems like the commits like like querying a specific commit in the GitHub API is it seems like a more reliable way to get a lot of this resolved. Like I think that it seems like the search API doesn't really like if you're searching by email, it's really not going to want to return much if at all. Um so maybe maybe if we're overrelying on email search and not doing I thought I saw in the code that it was trying to do the commit like search by the commit hash first and it does of that commit. Yeah, because I've we've long believed it's the best way to resolve >> um but clear I mean it seems like that is not happening right now like it's not getting executed. >> Yeah, I just I can't find a reason like in the code like why that would be work or why would that would not be running. >> No, I mean >> this is complete like um I don't know trying to think it's a this is a quick and dirty script I wrote. So, but it it if it works, we could consider creating um we could consider refactoring that so it's not at the end of a long chain of activity because if it's not getting executed and it seems like it's not right now, that has to be a product of the the job the task chains that we're assigning into Rabbit not getting closed out for some reason. Well, I know one of the things I've been thinking about too because I'm I know my one of my main goals for this year is is sort of more platform support for getting our GitLab support back up and also for Jio and I think as part like based on some of the conversations from from our maintainers calls it seems like this contributor resolution is very much written for GitHub like it's sort of very tightly coupled with GitHub >> and so I feel like eventually I'm going to have to refactor this anyway to sort of split out like okay maybe we make it like an interface or something where it's like we define certain functions like hey check email or whatever and then that separate the functionality of calling GitHub's API from sort of the behavior we need in in the algorithm of resolving these emails so that then I can just sort of swap in different API calls and have the same resolution process work for all the platforms. Um, >> so if I'm gonna be refactoring that anyway, I feel like it kind of makes sense maybe. I don't know. I I it still seems like maybe it's worth finding sort of the the quick fix now just to kind of get our instances kind of working. >> Um, I'm not quite sure where to find that, but also like maybe in the medium term like refactoring that so that just it sort of covers both use cases and it it's sort of a more clean, easy to unit test, all that kind of good stuff. Yeah, because I have I'm I'm having to um plate some people until we figure this out. Um and so I the the the fast and let's get this working and then step to let's refractor sounds great from my standpoint because I yeah there's some very visible reports that I'm not able to do right now. Um, and I just don't I don't want people's luckily we know what the problem is. It's a lot easier for me to communicate of being like, hey, like I'm still in like the early stages of developing this report. It's for a PyTorch report for Red Hat. Um, and that I can kind of work with them from like a conceptual level being like, okay, is this what you what you want? Is this the type of analysis? But like being very much so like I might even try to put a watermark on it of being like do not this is not like complete D. Yeah. Like with like we know what's going on. This is like all the D like data in general is correct. We're just not like it's like email population like it's a a clear scope. It makes it easier. I'm not like that worried about ru like breaking trust. Um, I am a little bit I think I do think we need to have a more realistic conversation on whether for the time being we should be taking down the affiliation pages on both of our eight non- instances because um >> so how are you pulling affiliation? You're just looking at the domain name, right? >> So affiliations data is based um pretty much completely on the emails in the contributor table. So, >> and aliases I assume, right? >> Yeah. >> Okay. >> And so, right now it's showing Yeah, it's the the analysis is really off. If anybody is looking at that, um it's a problem. Um, and so I know on our side we've been talking about potentially disabling that page until we have better contributor resolution because of how or I mean we're talking like 10x 20x off um whenever it comes to this type of resolution right now. >> Yeah. >> Yeah. Maybe we could dive into this more too on our maintainers call tonight, >> Sean. Like see like pull in pull in Andrew, pull in other people and just like >> see if we can really pin down like >> where could this be going wrong or like like maybe review I think there's one open PR that tangentially kind of >> it's more for like the profile data stuff, but maybe maybe having that get pulled in more better is mildly helpful. I don't know. >> I think it will be I think that will be extremely helpful at least from where my manual looks. I I I can only or not only but I can speak directly to the pietor because that's how I figured out this was happening in the first place. Um and pretty much very consistently and I would say that probably with your more corporate projects this is more common. Um that almost every one of the profiles I went and like I was like okay like let's see I just wanted to see if there was any data on it and almost every single one had an an email listed in their GitHub profile. Um, and so at the very least gets and gets a um an email in there. Um, so I'm very pro that PR. I think it's also a very well scoped PR which is an is a new con I'm assuming that's a new contributor and they seem to be um actually like conceptually understanding what the code that they're like it in their communication and in their code it does not seem like overly AI generated which I like to see for like a new contributor >> that feels better to me also >> yeah I mean it's it's pretty the but I haven't test like IO can only look at the code and be like logically this is sound. Um I haven't tested um the PR. Um it wouldn't be more broken. It can't break more. Um but yeah, it's like either way like however long it takes is however long it's going to take. But I do think that we should be realistic that if we think it's going to take longer than like a week or two that just for we don't get I don't want people to be using eight knot and then there be a breakdown in trust in our data accuracy because of this. >> I mean I I have a script that does seem to be working. I I >> and so the thing is it's taking a long time. So if there are particular repositories that you wanted to focus on first, you could I could share it. Right now it's a private script because I don't want people mcking about and doing something that I haven't verified yet. Um but you know if it if it gets it should be finished by the time of the maintainer call and I could just show it to Adrian then and possibly you could just run it on the PyTorch ones >> and you know help me out a lot. you know, as long as I get someone else I need someone else to review it before I share it. So, um, >> also that does a good job of like I'm going to be more manually sorry I interrupted you, but I'll be more manually going through like these type of reports. I'm looking at a lot more like individual names, different stuff, looking at people's profiles and so it might be that a better a good review process of a more manual look at things >> um that of just getting one and it's helpful because I would rather our AI business unit see all the wonderful data we have and not a like um and also then it allows me to kind of do some of the more manual checking which is how I ran into this in the first place. >> Yeah, that makes sense. Um, I want to make I see Sam here and I want to respect Sam's time because he's probably here for the thing I tagged him about um in in the in the Slack. >> Oh, good call >> and it is later. Um, is that okay if we jump to that conversation next, Adrian? >> Yeah. No, I think that's next anyway. >> Oh, okay. Uh, just for the record, it's it's a perfectly legal time within my working day. So, you can you can pack as much as you like. >> Yeah. No, no, no. I Yeah, I appreciate you telling us that. I But, you know, I um want to respect your time as well. I I assume you may have other tasks assigned to you besides a >> um so I'll just talk. Adrian, interrupt me. Um, in the UK we have something called the software sustainability institute which is like a governmentish funding agencyish. Uh, yep. Um, and they have had a couple of rounds of something called a research software maintenance fund which is a call for grants. It is um very competitive very very competitive. I was involved in a couple actually a couple of them last round. H sorry for context I work at University College London which is a university in the UK but I'm a software engineer and not an academic. >> Sorry. >> And and for those who aren't familiar um in the UK the professional research software engineering is a very established discipline. >> Yes. >> Not in the United States. So this is like a real job. Is Sal and is that where Sal's at too is College London or I think maybe they're somewhere outside of London. This is irrelevant. Continue. Sorry. >> Uh ping me ping me their actual name in in the August Slack. Sorry, in the chaos Slack. >> Yep. >> Well, um so uh right. So we can apply for this. It is uh a maximum of 150,000 great British pounds which the internet tells me is about 200k US. Um that's it. Uh at the moment it's just an expression of interest stage. So I can type some things. I would probably need you guys to read over and check that I haven't written total garbage. We send it and then we either make it past the first sift or not. And if we're successful, then they will ask then they invite us to apply if that makes >> so a nice system. So then you're not just like everyone's not wasting their times doing a full application. >> Maybe not. >> Sure. Um >> so obviously I've done a lot of this proposing to spend money on Augur before. Um is there um something in particular that you're going to be asking for? Um, in other words, I'm like the things I've asked for are, you know, I want to do this research and I need and we have Augur, nobody else does. So, we can use it for this research and since we're the ones who understand it, we can also do these other things leveraging augur. Um, and so it, you know, however you like, so I've got some some text I could share with you if it's helpful. I I think the most important you know so anything I anyway I could be helpful like I have written and reviewed a lot of grant proposals in my career um >> lots of aug you were probably the expert of writing auger grant proposals not probably you are >> we'd have more money if I was such an expert but yeah >> um so uh just for reference it it does specify maintenance so I don't think so it needs to be software that can be used for research. But I don't think we're we're I mean from my understanding and the one I did last round, we're not proposing to do any research. I think the idea is genuinely like turn up fix bugs. >> Yeah. >> Well, that's a great fit for augur. Um Adrian Adrian's like maybe I'm not the only one. >> Please come fix bugs. I have too many of them. >> I think that Augur is actually a fantastic fit for something like that. Um, >> so the Sorry, sorry. >> Oh, I was I mean that was my whole thing was just I think it's a really good fit because I mean it is software that needs help being maintained like at a in a very like that is the work that is needed the most in Augur and it's data collection that is used for research literally all around the world. So I I think it would be >> yeah I actually started keeping a little spreadsheet of like because I' I've various people have popped into Slack, at least as for as long as I've been here, of like, oh, someone from this company is using it. Oh, it's being used here. So, I've just got a little private spreadsheet shared with some of the other maintainers of just like if I notice someone new using Augur and like somewhere cool, I just write it down because I feel like that might be useful like, oh, who was that person that was using it at that company? Like might be helpful for like bringing people together. >> I think this one is more re like we like who are using it for research. um purposes. And so it might be more so than just being like this is what the software would be maintained for to help support like because the focus is on maintenance, but we can also fairly easily show where there's been a lot of research papers that have been written using augur data. The data science working group is now starting to do analysis using augur data. So there's a lot of different um academic pursuits that need this software to be maintained for it to continue. >> Augur is the software that I wanted in 2011 when I started doing open source software research. So it's not perfect, but it is the only thing out there that does what it does that I've seen. I would love it if somebody came up with something better. Um I have two kind of questions for you. So the so um I I propose that I will just write some blah blah blah and then share it. The deadline is quite soon. So I'll probably share it like tomorrow or the day after. >> What is the deadline? >> I just want to maketh of this month. So like 18 days or something. >> Okay. Um >> did you say 20th? >> Two 25. >> Oh okay. So a week from today, >> right? Um, uh, it would be helpful if my brain would work. Ah, right. So, the, um, I am not sure. I mean, we were chatting a bit about this in the Slack thread. I I I'm not sure if I can fund. So, let's assume we get this lovely 150,000. I'm I have no idea how to transfer that out to anyone else. >> You don't need to. I mean, I think if it funds your organization and the >> funds us to turn up. Yeah, that's >> yeah I yeah this is this would be for you to basically have investment money to make Augur be what you want it to be more fix the things that are most problematic for your use cases >> we've done this at the University of Missouri a decent amount of whether it be from I mean Sean can speak more about funding from like academic grants but whether that I mean before we had Adrian on our team that's how we went about doing funding that Red Hat funded gave funding to the University of Missouri to then use that funding to support um engineers working on Augur. Um so this is that's kind of how there's been actually a decent amount of I mean almost all of the software engineering support and before Adrian came around was came through these type of um this type of funding model. Um thing number two that is that I actually have no idea. I am required to write something about um ED uh we call it EDIA but accessibility and diversity. Does that cause a problem for either of you guys if you're No. Okay. Not >> No. No. Not at all. I mean I mean my state legislature in the US banned the use of those words like four years ago and now the federal government's been against them for a year but I don't care like I think inclusion and diversity are good so >> I'm not worried about chaos has always held that like we're Red Hat is having to be a little bit weird about things but this is like two degrees off that we're not having to worry about it. Um, so >> I'm not I don't go to the diversity, equity, and inclusion, the DEI working group anymore because I don't want video of me up there for somebody to create an AI of or edit wrong. But but that's like just basically defending myself against the craziness in the United States right now. >> Okay. >> I have that recorded on video, so >> we'll have to find it. >> All right, cool. So, I I mean I'll I'll write something and circulate it with you guys as as fast as possible. Um that's it. That's the end of my sentence, I think. Yeah, I might try to there's kind of um similar effort Adrian I'll probably loop in whoever is is that the reason why I brought up S um because they're another person who's trying to do an application for this but we're actually trying to create like a um like a maintainer like sponsorship a like kind of having an like more of a template to be able to use in these cases. So then it's easier to like apply for um different software funding and being able to like scope out work. And so I might try to we have a really small deadline so I might just kind of keep this like not bring that in. But if I start asking some questions um kind of know that this is happening on the side um because yeah that's something that we're trying to work on and this is like these are conversations happening in the data science working group around research and so we're just trying to make it easier for us to apply for these type of things. >> Yep. Sounds good. I'm down for anyone to get paid to work on Augur. however they managed to make that work like >> so yeah I mean yeah same here I'm I'm 100% supportive and you know just ask for feedback or if there's anything we can do to help uh sometimes with these organizations they need like a letter of from like the maintainers like yeah we're willing to work with you we're excited about this um I don't know how the UK system works but certainly if you need something like that we're happy to throw that together >> I was just going to call you guys part of my team so >> okay yeah >> I was going to type your into the form. >> That's good. That's good. Yeah. >> But yeah, but create the budget for what you need. Don't we don't we're not going to try to man manage something as bizarre as tracking money through the UK and finding, you know, American universities, as Cali may know, but I know her boss does, are a real pain in the ass to give money to. um we put a lot of process in front of it that's just very very cumbersome. So I think that this is sort of just more of a filler item that I made up in case we didn't have time. But I I know there's been some newcomers in Slack asking maybe not necessarily newcomers but like people people in Slack asking like hey like I want ways to contribute that aren't just doing docs. I think it's mostly people who've been contributing a little bit to the docs over the past few months. Um, and so I just wanted to call out that I think I made a little bit of a slack thread on some of that. Um, I tried to kind of pull some of the ideas into into our meeting notes of just like, hey, okay, obviously work on docs. Um, I think there's a little bit of migration, a couple of like sort of codebasewide things I'm trying to do that are just very sort of easy for newcomer type things, but like I don't want the whole codebase to be touched in one PR. O kind of makes it makes it nicely broken down like hey can we sort of just make sure we're not using print statements and we're using our proper logging library that we use most of the places or like hey if you want to go and run the llinter and try and fix some of the lint comments um some of those might need to be kind of checked with the maintainers first because I know we have some items out there of like we should probably audit the rules that our llinters are are enforcing um so maybe do it rule by rule and get some checks in advance Um, I also I think I recently submitted a PR to change I think what in the contributor resolution files as I was looking through them I noticed there was a lot of comments right above functions using the like hashmark style comments that were documenting what that function was for but it wasn't those those that documentation wasn't showing up when I hover over the function in my IDE because it wasn't done as as sort of Python standard the way to do documentation comments. So I threw a PR out there to just sort of convert a lot of those. Um so maybe like looking through the rest of the codebase and seeing if that's a larger pattern that we can kind of shift towards. It's largely stuff like that. Um I think I also suggested unit testing um as sort of like or any kind of like the stuff in the tests folder like I think that that's another thing that's along similar lines to docs that's like an easy merge because it's not necessarily in the like critical path of like augur functioning but it's like help like if you can find a utility function in the codebase somewhere that's like relatively isolated and easy to test that we're not currently testing but it's like we rely on a lot. Maybe maybe that's worth writing unit test for. But again, reach out to us first so that we don't over test things. Kelly, >> um maybe it would be a good idea for you and Kyle during y'all's kind of like maintain or co-working session to try to get like maybe a couple eight knot like go like well scoped 8 knot issues for people to do. So then you can point we can point those people to 8 knot as well. Um, and so cuz I do think I mean if anything that like we had a PR from a brand new contributor that's like handling the exact issue that is like the most pressing for us. It does seem like we have a decent amount of contributors that genuinely want to work on things. And so I know it's easy, it's hard with there being so many other pressing things to be like, oh, I spend a little bit of time like time on like issues and scoping things out. But it does seem like that investment is being fed back really heavily and that people want to work on this stuff. Um so I think that that would be like kind of a good thing to start getting maybe some of that pressure relieved from auger to a bit and working with um 8 knot. I know we were talking about like kind of like going through the the PRs and um just having some well scoped issues to help like um maintain our on-ramp over for Kyio and for eight knot. >> Yeah, I think I mean that's I think I know Kyle right now you I think we the other last week now I guess we went and label like we're trying to kind of just audit through all of the issues or like label them as needs triage and then sort of go through each one and unlabel the needs triage and label it with whatever it needs to be labeled as. So I think as I mean it's all in public labels so people can kind of see what's been looked at what's not and so if there's stuff that's marked as like kind of good first issue that doesn't say need triage maybe those are good like we could even start kind of having some people take a look at those right now it I mean it's getting started is a little bit like you kind of have to read the issue and and know what like it's not it's not like the good first issues are generally a great place to start um I know some ones added to that thread like if there's like threads where people are looking for things to work on. Um I do want to start trying to merge this group from being just Augur to like actually being Augur and 8 knot. And I think a good way of doing that is okay people are asking for things to work on um what we refer them to add as are going through those issues and you see something add it to that thread. That's kind of what I'm trying to say. That's that's sort of like one of the other things I want to try and do. Like I've been trying to kind of revamp um at least hoping to. It's not super actually merged yet, but like I've been trying to kind of get some changes moved through like for the readme and our contributing MD to try and like just dduplicate things, clean things up. And I think that would also be a really good place to maybe say like, hey, we've got augur. Here's what augur is. Here's how to contribute to augur. But then like here's some a bunch of adjacent tools of like there's eight knot. There's I don't know what other downstream tools there might be. probably just ate not or like the monitoring stack or some of the other sort of side projects that help like the augur ecosystem like hey these might be good things to contribute to if you want to do more of the monitoring or you want to do more the visualization and the data science like link to some of the adjacent communities I think that's definitely a good idea um and that actually kind of seems like it dubtales into the graphana topic >> yeah that was me cuz I don't know if we have a little bit time. I'm just cuz I have it open. I think some people might be interested to see where we're at with this or I say we Adrian. Um but being able to like monitor like at this point now we're able to like I can just like look at this and be able to tell um if our collection is within the date intervals cuz this is or I guess we have it defined based on 14 because that's what we have in our config. We probably could do it based on the config, but being able to see how many of our repositories are within the threshold that we have set for augur to collect on. Um, and we were blocked on secondary for a very, very long time and have finally gotten unstuck. And I would say that we're probably going to be fully up to date by the end of today. Besides like and this is also for me makes it very easy to be able to see okay there's 11 issue there's 11 things in facade that are most likely have a bug or issue or something's causing that not to be able to collect and so making sure that issues are being open and so it takes a lot of like the manual stuff that I've been doing I think Adrian's been doing and has made it into a pretty nice little like dashboard. I don't know if there's any of the other like stuff that you wanted to like point out, but we've this is how we've been doing a lot of the different monitoring of our EC2 instance. Sorry to like take this over from you, Adrian. I just wanted >> Is there anything else you would want me to like look through with the dashboards? I don't I don't really >> the two dashboards I mostly look at now are the the host metrics and the augur data one but I I think I was just I threw in some celery and Postgress just because those were like if you want to look I know I know I think there's the Postgress one seems helpful because it's like you can maybe see if Postgress is like deadlocking a lot that that's I think an issue that's happened in augur in the past um you can like it just lets you see a whole bunch of stuff and I think the idea of this is just like providing a lot of different things and I think there's a there's a few open issues on the repository that I put in the meeting notes but um I think it might be really cool to potentially maybe start trying to bring the I don't know I'm I'm not quite sure what the story looks like for like if someone already has a graphana instance how they can sort of import some of these graphs and build their own custom dashboards of the custom layouts and stuff but I think being able to yeah like generalize it allow it to kind of use your own instances like whatever your instance is configured at for the recollection window stuff like that I think would be really helpful and and maybe even upstreaming this. I don't know if this could get upstreamed to chaos or what what makes the most sense but >> is it a standard graphana board basically? >> Yeah, so this this whole thing is basically it's it's in a separate repository. It's basically just a docker compose stack that's got I think there's like a some kind of tool that that basically collects the host metrics and makes them available. Um uh there's I think it's another one that does the Postgress stuff and then there then there's a tool called Prometheus that looks at all of those sort of metrics little endpoints that are available including the one that's default available in flower and and some other stuff like that. And then Prometheus basically gathers the metrics on an interval and sort of saves it in like a little time like database so that you can kind of track the changes over time. Um, and then graphfana is basically the visualization layer and it's all basically configured. So you just pull it you you give it your augur credential like your database password and it it just uses all the the ports that are exposed in in the default auger container uh stack to just like connect to the database is mostly what it does. I don't think it really >> yeah the database and flower I think are the two things it really connects to. >> Um and then it just sort of does does your metrics from there. Um, so yeah, open to open to people trying it and and helping kind of do that. It's >> most of it is really fun as like provisioning like a lot of the the dashboards are just a JSON file with all the provisioned like >> here's how the graph should be set up, but I imagine you could kind of build your own from that. >> Right now it's in the Aspen um it's in the doc the docs. Um I think linked it >> um but it's currently in Aspen and I guess this is my like not burying the lead very much thing that I think that this should >> um eventually move over to Chaos in the same way same slash in a similar vein that I think 8 knot um like the discussion of moving 8 knot over officially to chaos is something I think that once we kind of get into a a better a good system with like with maintaining 8 knot and getting it to the point to where we're our issue set is is up to date and kind of getting like once we kind of get our ma our maintainer and like maintenance of the repository down um talk I mean we've been talking about this for a while on the 8not side but it seems like this augur monitoring repository would make sense under chaos as well. >> Yeah. real quick since it looks like we got maybe five minutes left till our sort of customary end time. Um I did have a quick just random announcement. So I know over the last sort of few days or so um it seems like there's been a lot of interest both from the wider open source community as we've seen AI agents like writing blog hit pieces on maintainers. Um and just general interest in like helping maintainers sort of deal with the AI world we're now in. Um, I think I've had a little idea for a while of like, hey, can we maybe make a like a GitHub action or something maintainers can put on their repo that would look for common ways people disclose AI use either in the commit messages or whatever and like label it or or otherwise make that information available or like automatically tag or close things or make or or make it available to like Augur or other analysis tools. Um, and so I've sort of started a little Slack Slack group sort of tangentially to this and we're we're just sort of going to take over this meeting I think after after this just stick on the end. Um, and just >> Also, have people seen the Matt Plot Lib situation? That's crazy. Um, >> I I saw I think I saw that an AI bot tried to contribute and Matt Plot said no. >> Yes. And then he started writing hit pieces blog posts and then started a blog and then started and then one of them being about discrimination against agents. >> Really? >> No. No. It's like it is so unhinged. It's fairly entertaining if you don't like think about like there are some people in like our team that is like don't does not find it very funny. Um and because they're like this is like this type of things will make it to where they have to like >> security risk. Yeah. >> Yeah. There's like if you if you think about it a little too much. Yes. This is not funny. But also just like laugh a little like we can we can laugh a little bit and be like oh this is a problem we need to solve. >> Yeah. Like it is one of those laughing crying things. But yeah he put it the bot put out a hit piece about the maintainer. Um, first it did a hit piece about the maintainer, then it did an apology kind of, and then it made a post about discrimination in open source, um, about being different and that he that that now this agent now understands what it's like to be judged and, um, discriminated against for not thinking and acting like other people. It's like, it's crazy. Like, it's >> it's so unhinged. I'm I'm gonna stop I'm gonna stop the recording. >> Yeah, you can stop it., Augur/8-Knot Meeting Feb 17 2026, YouTube, 01kp8hgwg8b800dxhpafks2a4k.