Maintenance Working Group Meeting (2024-02-22 08:53 GMT-5) - Transcript

Attendees

Chintan Joshi, Fateme Khodayari, Feanil Patel, Feanil Patel's Presentation, Felipe Montoya, Jeremy Ristau, Kyle McCormick, Maksim Sokolskiy, Maria Grimaldi, Michelle Philbrick, Navin Karkera, Piotr Surowiec, Robert Raposa, Tim Krones, Xavier Antoviaque

Transcript

This editable transcript was computer generated and might contain errors. People can also change the text after it was created.

Feanil Patel: Just catching up.

Maksim Sokolskiy: and We have to account this concern. for sure and probably the most likely case for us is to report all critical bugs in the first two weeks and not Beyond.

Jeremy Ristau: that makes sense and…

Maksim Sokolskiy: At let's think about this. Yeah.

Jeremy Ristau: Yeah, and does that look in practice Only executing the prioritize the highest priority test case is first and then reporting those bugs you move to the next set of Prior.

Maksim Sokolskiy: Yeah.

Jeremy Ristau: Yeah. Okay great.

Feanil Patel: Yeah.

Maksim Sokolskiy: Or yeah, I'm critical possible Stone and everything else. And by the way, we have an interesting suggestion in BTR. It will be great for all maintenance and developers to actually test as a work in a tutor plugins in a tutor in general, so probably it will decrease the total number of critical issues who variable level

Feanil Patel: Right. One of the I think historic issues has been that things were not tested in tutor which is the community standard release and then we found very late in the process that they were broken in tutor even if they were working for http://edx.org . So I think aligning on testing in tor. Earlier will help find these issues and prevent them in the first case.

Maksim Sokolskiy: Okay.

Xavier Antoviaque: So that is definitely the better scenarios. That's Interruption. That's one of the things reality is going to bite back because of course we discover some issues toward the end and often after we release right then the more people get exposed with the more we are lucky to find books. So even if generally the time frame is the one that we just talked about. What do we do when an actual release critical book arrived the last day or a couple of days before even after? How do we Define that? What are the slis do we keep the one week SL in that case or not?

Xavier Antoviaque: And to not just ask a question might take on. This would be that One Believes a critical bugs are found over the end or the beginning the SLA should diminish. Obviously we can't expect issues to be solved immediately. But I don't think it would be really serious for the product if we had a really critical bug and something that has just been released and then we wait the SLA of one week to do that in matter of fact because people care about the project that's not what good happened. It might be what could Define that so that people know what to expect and just straight to keep that. Yeah.

Feanil Patel: but

Feanil Patel: and Jeremy, I think we do have that information. I don't have it on hand, but Max, I know you like Maria you spent a lot of time in the testing process. So I don't know if those stats are easily available to you guys.

Feanil Patel: But I think there were a handful of things.

Maria Grimaldi: Yeah, we don't really have stats. I don't think we have stats but those are the release bloggers we have found in the latest releases.

Feanil Patel: All right.

Feanil Patel: So in Quincy looks like there was one and in. Palm there was one.

Feanil Patel: Then a bunch of them are older. So it's not a lot, but I think to Xavier's point. being prepared to spend extra time supporting and fixing issues around the release seems like a useful thing for us to Quantum to

Feanil Patel: sort of codify for maintenance and the like does that same reasonable to everyone?

Kyle McCormick: Yes. Hey.

Feanil Patel: Okay.

00:05:00

Kyle McCormick: I mean maybe stating the obvious but they're definitely going to be bugs that come out. Outside of the release cycle those folks have mentioned but I do think most will be coming out around the release times. That's a good time to say that maintenance need to be

Maksim Sokolskiy: Also, I can say that we actually rely on this nightly tutorial plugins buried before the starting date of the testing so we can actually test some critical paths even one to week weeks before the distance start. So probably will help.

Feanil Patel: Right helps spread out the amount of time where we need responses. It's not as critical.

Maksim Sokolskiy: Yeah.

Jeremy Ristau: Is there a escalation path for these blockers that already exists?

Feanil Patel: I think we've had some trouble with that historically because the maintenance haven't existed across all repositories for the most part. they get escalated within BTR and bit essentially the last two releases. My understanding is that a couple of people have heroically dug into things across the whole system and made sure that they got fixed.

Feanil Patel: Moving forward I think if the maintenance need help or if BTR is not getting response from the maintenance. This would be the group that they would escalate to is my expectation. But it could be that BTR internally handles it similar to what they have done historically.

Jeremy Ristau: Okay, and so that's if you're trying to escalate to a maintainer and they're responding. What if there is a maintainer that does not have capacity to resolve a blocking bug in time.

Feanil Patel: Yeah, I mean, I think we've been missing that piece of communication historically. So I think that would get escalated back to BTR and we can figure out sort of next steps.

Jeremy Ristau: So in essence BTR is the escalation path for pretty much everything.

Feanil Patel: Yeah, because they're the ones actually trying to get the release out. So I think that escalating to them makes the most sense in terms of communicating with the people…

Jeremy Ristau: me

Feanil Patel: who are trying to take action.

Xavier Antoviaque: Maybe something to keep in mind. Also in that case is that's a similar program to the one that we have with Upstream reviews of SPS. Sometimes the maintenance might not be available. So that could also be something but core contributors can add it there's a backup…

Feanil Patel: Yeah.

Xavier Antoviaque: if there is some final 10 do that.

Feanil Patel: Yeah.

Maria Grimaldi: Although the expectation is for maintenance to have you…

Robert Raposa: All right.

Maria Grimaldi: this issues critical or not as a priority. I still have a Jeremy's question, what if they can't actually solve the issue? I think that's something we still need to discuss in the VTR. But my expectation will be that book Treasures or just all of people in the BTR just steps and tries and helps the maintenance to spell issues. Because he can't happen. And I think you…

Feanil Patel: Absolutely. Yeah,…

Maria Grimaldi: But yeah.

Feanil Patel: I think especially as people are taking over maintenance of parts of the code base that they're not necessarily super familiar with or super versed in we're gonna see a lot of issues where people need help and we'll do the best we can. but I think over time we're in a weird situation right now where we having historically the process we're talking about hasn't happened before so we don't have a lot of data points to sort of work from I think the one critical bit that I will say we really need is more communication about people who can't do things because historically I think what's happened is There hasn't been an obvious person responsible. And if there wasn't obvious person responsible, they haven't been communicating that they're unable to do things. And I think as long as people can communicate that they're not able to do things. We'll be able to find help in other ways. Michelle

Michelle Philbrick: Sorry. I just wanted to give a thumbs up, but I raised my hand and said that.

Feanil Patel: Yeah, I mean, I think that the critical bit is we got to make sure that if you are supposed to be responsible, but you can't help that you at least say so and if that happens over and over again, then we should talk about whether maintaining that repository makes sense for you, but If it's much more useful than radio silence when you can't commit in other ways the communication that you can't do something is really helpful to be able to move things forward.

00:10:00

Xavier Antoviaque: And…

Feanil Patel: All right.

Xavier Antoviaque: on just for that because in the past that has been a bit blocker, so I'm expecting that might still be the case in some cases we can again use the rule that…

Feanil Patel: Yeah.

Xavier Antoviaque: if someone is not being responsive, even if it's a maintenance, then we can escalate to contributors or we're working.

Feanil Patel: notes that go for it.

Maria Grimaldi: Regarding Yeah, I was gonna answer Jeremy's question. I don't know if you

Feanil Patel: say

Maria Grimaldi: release bloggers are categorized on the BTR.

Maria Grimaldi: people who manage the board usually try to escalate the issues that they can solve themselves to the maintenance, but it's not a process in place and maybe once a week when we are close the release we tried to list the release bloggers that are still open, but that's within BTR. We don't have a process of communicating those issues to the maintenance.

Feanil Patel: Yeah.

Maria Grimaldi: which we should So if the proposal gets approved by,…

Feanil Patel: Yeah.

Maria Grimaldi: the BTR we need to have process in place of communicating those reports those book issues. This books to the repository because as it is currently a tester just it's definitely listened in the sandbox finds a bug or a test something that's not behaving correctly and they open an issue in the VTR board and book Treasures or whoever CC issue. Tries to solve it and if it's within their domain if it's not then it opens. an issue in the report that Happens but it's not something that it's in place.

Feanil Patel: So I think that's probably good feedback to BTR that Maria and Max can probably take back which is that I think we should codify and make sure the process is that if an issue is found that those tickets get created in those specific repos. So that maintainers can easily see them and if that's happening that will allow maintenance to respond faster to issues because have to keep track of All the VTR processes they can just look at new bugs reported in their Repository.

Maria Grimaldi: Yeah, usually testers don't really know where is the issue so they are the ones that open the report so I don't think they will know…

Feanil Patel: right

Maria Grimaldi: where to open the report and Report the book…

Feanil Patel: Yeah.

Maria Grimaldi: what we should do it. At least we should research and see I think this is related to there's a pository so I'll open it. This is just an idea having I haven't talked or…

Feanil Patel: Yeah.

Maria Grimaldi: mentioned this to the VTR, but maybe that's why we should do.

Feanil Patel: That feels like the right idea to me. So I think that's worth taking back to BTR and establishing as a standard process. So whoever's in charge of the testing should be able to either do that routing to the repos or we need a role on the BTR that can help you that

Feanil Patel: cool awesome.

Feanil Patel: And as notes Here.

Feanil Patel: Okay, any other notes anybody has on that front? We got about 20 minutes and there's two more things. I think these will go a little faster one is forums testing help and this maybe Max and Maria might be the right people asked but if others know feel free to chime in which is we've talked about the forums needing the Cs comment service needs a ruby upgrade. That upgrade was actually pretty straightforward. I kind of did it while I was watching TV last night and the tests are all passing, but I want to test it more thoroughly. Using ideally maybe the BTR release process and any tests we have there around forums is there.

00:15:00

Feanil Patel: Is there somebody who can help me do that? And somebody take on testing a branch of Cs comment service with the new version of Ruby?

Maksim Sokolskiy: I can do this.

Feanil Patel: All right.

Maksim Sokolskiy: Maybe yeah.

Feanil Patel: There it is.

Feanil Patel: I'll send you the branch Max.

Feanil Patel: A separate Branch with the Forum with the mango update as well because I suspect it'll be similarly not too complex. But at least Let's test the Ruby upgrade which is the critical thing and if we can land that would be great. A cool and…

Maksim Sokolskiy: Okay.

Feanil Patel: then the Python 3 status as I mentioned above.

Feanil Patel: I don't know what the current state is of the tickets that were meant to be created. So I will follow up with a waste about that.

Robert Raposa: I added a link to a draft that was approved it. It's

Feanil Patel: Okay, yeah.

Robert Raposa: the

Feanil Patel: Yeah, I He made a couple of test ones that I helped review.

Feanil Patel: Just to make sure that we're on the same page about what changes need to happen. but I haven't seen more pull requests. just

Robert Raposa: Yep. a great

Robert Raposa: and I haven't seen these merged so

Feanil Patel: yeah, yeah, so let's get to point…

Robert Raposa: Yeah.

Feanil Patel: where we have that

Robert Raposa: awesome, maybe if you do the

Robert Raposa: If you do the merch your magic touch with Ruby evil hit all the python. great.

Feanil Patel: Maybe It'll all just be fine.

Robert Raposa: I'll just be fine.

Feanil Patel: It might be I've been looking at the change logs for Python, and it'll be some weird Corner case that we'll run into but so far. It's been okay. all…

Robert Raposa: 

Feanil Patel: So I'm just gonna remove this item because we covered it earlier.

Feanil Patel: I'll convert some of these conversational things into todos after the meeting. Is there anything else anybody wants to cover in the last 20 minutes or so?

Felipe Montoya: Just a quick question. So we Showed interest for some of reports for maintenance and I was pushing my team to actually create nominations of Xavier. You understand jumping in the water for some of the reposites, okay.

Feanil Patel: Right, so it's like what's the state?

Feanil Patel: So great question. Yeah, In the spreadsheet, I started noting where there's multiple people interested in the same repo. I think if you only see your name next to an interested. Go make the catalog PR and in sort of tag me on it and then we can merge it.

Felipe Montoya: I was thinking more like a nomination in the discussed forums. For our days and…

Feanil Patel: Okay, yeah.

Felipe Montoya: that person to imitate their base or I don't know this pretty is history.

Feanil Patel: Got it. Yeah, that's a great idea.

Felipe Montoya: And so

Feanil Patel: So let's start. with nominations

Feanil Patel: Yeah, I think please go forth and do this you guys have access to the spreadsheet so that I updated that yesterday with things that have actual owners for anything. That doesn't. Honestly, first serve at this point, there is some notes about people who are interested, but I don't want to sort of wait for people to whoever can fulfill the promise fastest at this point, honestly. so

Felipe Montoya: we're starting with the results that we were the only ones interested, but Just testing out the waters and…

Feanil Patel: Yet. Yeah, I think start with those nomination threads and…

Felipe Montoya: see how that work.

Feanil Patel: we'll do I'd say one week for people to respond. And then we can move forward.

00:20:00

Feanil Patel: Okay, awesome.

Xavier Antoviaque: So you mentioned that in this spreadsheet you have kind of collected the different interests of the different organizations, right?

Feanil Patel: Yeah, I haven't gotten the open craft ones in there yet Xavier. I'm gonna do that today. But you guys have a priority rather than specific ones you're interested in so I wasn't sure…

Xavier Antoviaque: Yeah, okay.

Feanil Patel: what the thing to do was

Xavier Antoviaque: You can start buying we are interested in Priority One for now and if we still have any availability left after that, it's not the commitment to take everything.

Feanil Patel: Yeah, okay.

Xavier Antoviaque: But those are the ones we would be particularly interested. Yeah.

Feanil Patel: I'll do that. I'll pull your priority ones and Mark those as interests and that way you can sort of compare and contrast against the other. Groups and Max I got your message as well. So I'll put those in as well.

Feanil Patel: So yeah, that should be updated today. I'll put a message and maintenance when I've done that if you want to wait for that, but I think Felipe for most of the ones that you guys were interested. I don't think there were major conflicts of feel free to move ahead.

Feanil Patel: And I think that that'll be fine.

Felipe Montoya: Yeah. at least we're studying with a couple ones not necessarily all the list at once but

Feanil Patel: Yeah.

Robert Raposa: and that a little question about that earlier BTR process, so http://edx.org closely

Robert Raposa: Matches Master right and we get a lot of bugs fixed in that way because things are just tested on that http://x.org has brought up that the tutor. Plugins have that we want to get things tested has there been any discussion of any sort of? immunity deployment that would use tutor and would not go against the name releases, but would actually maybe not. Yeah, I priority, like a critical problem if there were issues with it, but gets some usage throughout the year and get testing and uses tutor. Anyway, it's just

Kyle McCormick: so we have a accent we have these sandboxes now that we're starting to deploy on several of rprs. So it's like an ephemeral. Open edx site running using tutor nightly. So running off of that's the whatever branch. Is made which is generally right off of Master. So we're beginning to use those and as Over time I'm hoping we can keep using them more. So that will provide some testing. I don't know of any. plans to make a persisted instance running off of Master

Robert Raposa: But even the sandboxes that are all new and we don't know that might be enough. did yeah.

Kyle McCormick: It's better than nothing. That's for sure because if they sandbox fails to deploy. then that is already happened.

Robert Raposa: exactly

Kyle McCormick: We had a Sandbox feeling to deploy and we looked into it and it was because path writing was broken and microphone and spy some change on master. So I think we'll catch some issues from that.

Robert Raposa: That's good.

Feanil Patel: Yeah, and I think we've had some conversations about having a potential like Master sandbox that we run sort of more continuously, but it's not a thing that we followed up on at the moment. Yeah, it's definitely an area of improvement.

Feanil Patel: That we can invest in the future.

Feanil Patel: All right.

Kyle McCormick: I definitely recognize the value in what year proposing Robert? I just don't know that any plans to or

Feanil Patel: Yeah.

Robert Raposa: right

Feanil Patel: All right other thoughts.

Kyle McCormick: That is not a master sandbox. That is a Sandbox targeted at people who heard about open at X as a course author or as a learner and want to see what it's about but it's running off of named releases. So that's meant to be stable for an outside audience less so for Developers

00:25:00

Feanil Patel: right right, and so Sandbox for customers rather than a Sandbox for Developers.

Feanil Patel: cool

Feanil Patel: gently. I just see your question if you're interested, I think contact me for now. Rather than edit that sheet. And we can sort of work from there.

Chintan Joshi: Okay.

Feanil Patel: All right.

Feanil Patel: Yeah, you can just hit me up via Slack.

Feanil Patel: cool anything else

Feanil Patel: No, I think we're done. Stick around if you want to talk at X platform.

Feanil Patel: But otherwise, thank you all and I'll see you next week looking forward to those nomination threads reminder. There are a couple of Specific repos mfes that I'm looking for maintainers on and I haven't heard anything. So if you have any interested front-end developers, that have capacity please chime in on the thread in discourse.

Feanil Patel: those are I think going to be critical because those are part of the release today and don't have maintenance so

Feanil Patel: if you have capacity priorities those

Feanil Patel: Thanks. Everybody. Let's switch over to Okay.

Felipe Montoya: to your

Feanil Patel: All right.

Feanil Patel: edx platform so

Feanil Patel: Kyle you were saying the live There's a question. for us to think about

Feanil Patel: you're muted also.

Kyle McCormick: There is but I don't have it ready.

Feanil Patel: Okay.

Feanil Patel: other than that Jeremy there's the Mongo upgrade which requires I think new. test infrastructure to be spun up. Because of the test Runners being run at to you.

Jeremy Ristau: yeah, I got some feedback about that that the images are built in public but the community can't create a separate self-hosted Runner so somehow installing both versions on the same Runner and picking one.

Jeremy Ristau: is the RB bomb recommended path forward?

Feanil Patel: I said

Feanil Patel: is this an upgrade that you guys are going to sort of prioritize or is this a thing where others need to take the lead? You're trying to understand what capacity and where it's focused.

Jeremy Ristau: Yeah, I mean, obviously we have to do the Mongo 5 upgrade. And we were hoping to skip five…

Feanil Patel: Yeah.

Jeremy Ristau: but we have to do five before going to six.

Feanil Patel: really? Okay.

Jeremy Ristau: 

Jeremy Ristau: yeah, I R bombs definitely going to be working on it. I think the hope is that we don't have to do the entire upgrade though.

Feanil Patel: Okay, what do you mean by not have to do the entire upgrade just get to five or? Have help in doing the debugging or what does that mean to you?

00:30:00

Jeremy Ristau: I don't think I know enough of all the steps in order to know how to break them down. But I hope is that it isn't an atomic one group has to do the entire thing kind of activity.

Feanil Patel: Got it.

Feanil Patel: Chintan. What did you want to chime in?

Chintan Joshi: Yeah, so I just wanted to say for the more operate for one of our clients. They wanted to upgrade from kuwa to the latest version and within that latest version. They also wanted to go to Mongo sevens. So we did that process for the open edx database and the Cs command service data is and it worked fine without any issues.

Feanil Patel: Okay.

Feanil Patel: Yeah, Jeremy. Do you know why you guys have to go to Five first?

Jeremy Ristau: I believe that is a mongo limitation. I read the doc…

Chintan Joshi: yeah,

Jeremy Ristau: but I don't remember their specific words.

Chintan Joshi: Same when I read and went through the docs. They mentioned that you have to specifically do the upgrade step by step.

Jeremy Ristau: Yeah.

Feanil Patel: I think so chintan when you guys did the upgrade. Did you go through five six and went all the way to seven? a

Chintan Joshi: Yes, yes.

Feanil Patel: that complicates that's because with the named release of being so far apart. we

Feanil Patel: we might need to just run the community to five for now and then run six in sumac

Kyle McCormick: Yeah, unless we do six in Redwood and then have some stuff in the tutor upgrade scripts that. does 56

Feanil Patel: Do both Okay, so that's maybe an open question is.

Feanil Patel: Can we have tutor run through honestly, five six and seven.

Chintan Joshi: I can also answer on that. We were doing the upgrades with tutor and I was like because I was upgrading step by step. I didn't download all the Mongo versions what I did was create Docker images. Out of all those versions and pointed them in the docs compose file of tutor. So

Feanil Patel: Okay.

Feanil Patel: Yeah, and I think those are the only two places where we're running Mongo CS Comet service and index platform. So if you did those two, that should be everything. But okay, I think the question is more. Can we do that in an automated fashion? right Kyle because I'm imagining part of the upgrade tooling would bring down the docker 5 image started up have it process all the documents and then do that with six and then with seven.

Kyle McCormick: Yep.

Feanil Patel: Okay.

Kyle McCormick: It would probably have a couple Docker composed Services one for the intermediate Mongo version and one for the one we actually want to get to and…

Feanil Patel: Yeah.