Maintenance Working Group Meeting - 2024/11/14 08:59 EST - Transcript

Attendees

Adolfo Brandes, Feanil Patel, Feanil Patel's Presentation, Jeremy Ristau, Kyle McCormick, Robert Raposa, Sarina Canelake

Transcript

Feanil Patel: two Kyle. Now we'll get all this work done finally roots.

Kyle McCormick: Yep, I did it. Fun myself.

Feanil Patel: All right. I've just updated the support windows so that it has the new elastic search dates in Share my screen.

Feanil Patel: All righty. let's go through the previous to-dos real quick. follow up with Ed about code jail service. I did have some conversations with them. I think everybody's in alignment that that service should become part of the core. I just need to figure out sort of what the next steps in planning and change management for that looks but I think that's looking good. So, I think we can start thinking about sort of how to get rid of code jail as a thing that runs in the platform. this is maybe a conversation we should have on the tutor side as well. yes, I can.

Feanil Patel: 

Feanil Patel: Because if codeJL is not running if we want to get rid of the being able to run codegill on the same container as Z edex platform as a core feature, then I wonder if running the code gel container becomes more of a core requirement since without it you can't really run Kappa. Yeah. Yeah.

Robert Raposa: You're on mute, Maybe if you're talking

Kyle McCormick: Multiple devices. When we say core finale, you're talking about the reference product that product or whatever we're calling it these days.

Feanil Patel: On by default in tutor reference product I think.

Kyle McCormick: Yeah. Mhm.

Feanil Patel: 

Feanil Patel: which I know how much we love adding more containers to the default runtime of tutor but this seems like from a security perspective would be a sufficient enough improvement and…

Kyle McCormick: Yeah. Yeah,…

Feanil Patel: I don't think there's a world in which we don't run with kappa as a thing that is on by default.

Kyle McCormick: I mean we can run with Kappa but not advanced Kappa…

Feanil Patel: M okay.

Kyle McCormick: where CodeJill is supporting advanced Kappa. I don't think I could make that call as not being in the project working group…

Feanil Patel: Yeah. Yeah.

Kyle McCormick: but I think whatever the working group wants we can support code jail or…

Feanil Patel: Right. Okay.

Kyle McCormick: no code jail by default.

Feanil Patel: I'm just taking some notes.

Jeremy Ristau: 

Jeremy Ristau: So like to reach a decision on this.

Jeremy Ristau: This question will go to the product working group. Now Yep.

Feanil Patel: I think it will…

Feanil Patel: because it's in that vein of technical product decision where there's technical impact but also based on the decision it will change how the's default product is. So heat.

00:05:00

Kyle McCormick: Regardless of whether it's on by default, it does sound correct to have it that support in the open edex organization and the plugin be an official duder plugin.

Feanil Patel: And when I spoke with Felipe about it, I think he was pretty on board. I think we was just figuring out the mechanics of it. that feature and I'm going to talk to Mois who's the maintainer of the code gel repo right now. and I think also the maintainer of the code gel tutor plugin for edgy next. So hopefully it will come in with a maintainer that is already in our program. But again, I think we just have to figure out sort of what the actual mechanics of this looks like. But I don't think there's objections.

Feanil Patel: I'm realizing now that I actually have a lot of figuring out the mechanics of nontechnical code changes tasks on my list right now. all right. So, I think does anybody have anything else to say about the elastic search support windows I just updated. so this is now up to date.

Feanil Patel: The latest two versions don't have end of life dates except for the fact that the end of life date for version 8 is whatever is 18 months after when version 9 comes out and the end of life date for version seven when version 9 comes out is how elastic search does it. So it's nice and unpredictable which is fun. but it doesn't look like nine is coming out anytime soon based on their public information. So for now we're good with seven 17 and we could sort of keep working from that assumption.

Feanil Patel: I will say though there's movement to replace this with Melier search that I think is in development I don't have all of the details on this, but I've seen I know on the tutor side they're looking to have melee search be the default search in tutor. and so there have been updates to edex search to be able to support either

Feanil Patel: but it does I think given how annoying and complex elastic search is and how big it is in terms of infrastructure does either we need somebody who's maintaining the ES part of edex search or maybe we get rid of it to simplify our stack a bit. there is a nomination for an edex search maintainer. who will probably have a say in that but I assume that will go through a proper deer if it happens.

Jeremy Ristau: 

Jeremy Ristau: So, I align with everything you said and then the first sentence you typed feels like it doesn't align with that. all uses of elastic search are being replaced is in progress.

Feanil Patel: Yeah,…

Feanil Patel: let me make that more clear. Yeah. Yeah.

00:10:00

Jeremy Ristau: The expansion. Yep. Yep. Appreciate that.

Feanil Patel: Thank you. That is more accurate. let's see what else we got. And then I have not done this enabling cron ticket, but I'm going to try to do that this week.

Feanil Patel: And then chairman, has there been any movement on the deer for the old the studio replaced MFS at all?

Jeremy Ristau: 

Jeremy Ristau: No, no. what I would say is less than minimal coverage in authoring right now. So, we don't even have space to try and…

Feanil Patel: Okay, got it.

Jeremy Ristau: roll out pages and edge right now.

Feanil Patel: Okay. …

Jeremy Ristau: Yeah, I'll let you know when it changes.

Feanil Patel: hang in there. that sounds good. any other topics we want to cover today? Favor. Do you want to cover that? Kyle, I know you and Robert were speaking on a PR about favor. Are you guys aligned?

Kyle McCormick: Yeah. I'm working with Usama i. we're currently delaying merge till Monday at least. We've decided to do the non-breaking changes first, which works well for axe improvements and also gives TU a little more time. So things are chugging along there. I think we're aligned.

Feanil Patel: You know what? We should talk. This isn't upcoming maintenance, but the forum v2 stuff, I don't know if everybody here has been following. So, Adalfo, give a summary of forums v2 decision making. No, I'm just kidding. I'll do it. Adela has been doing this in the last three meetings. So,

Adolfo Brandes: 

Feanil Patel: 

Adolfo Brandes: Yeah this is my third iteration.

Adolfo Brandes: So basically what we decided to do as a group yesterday during the BTR meeting was for any new user of tutor the forums v2 code is going to be the only thing they can enable and it's going to run on a MySQL back end.

Adolfo Brandes: for existing users of a Twitter installation when they upgrade they're also going to be running forms the forums v2 code but the back end's going to remain in MongoDB because in conversation we identified that the biggest risk was the migration of the data from Mongo to my SQL given

Adolfo Brandes: that there's really way back right it's non-destructive the migration but once you're in my SQL the data is not being written to MongoDB at the same time so it's a diverging situation so there's really way to go back the other decision was that for the duration of the sumac cycle and the te cycle There's going to be a management command to migrate the data from MongoDB to my SQL. all of this is just about the forms, Just to make it clear, there's going to be a management command. but by teik, you're not going to be able to run the forums on MongoDB, right?

Adolfo Brandes: 

Adolfo Brandes: So people that want to try things out are going to have the sumac cycle to do so where they can theoretically switch from one to the other and test things…

Feanil Patel: 

Adolfo Brandes: but by te only my SQL back end is going to be available including in master right so I guess this is where people that run master prick their ears up so there's a bit of a question of when exactly that switch and master is going to happen. So I guess that sort of sums it up.

Feanil Patel: Right. Yeah.

00:15:00

Adolfo Brandes: Yeah. Go ahead.

Feanil Patel: Yeah. Yeah.

Kyle McCormick: Yeah, Adelfa, as much as you've gotten fantastic at explaining that really clearly and briefly, is that all on the deer ticket so that we have the paperwork to remove it to, make the breaking change between Sumac and Peak that everyone can refer to?

Adolfo Brandes: 

Adolfo Brandes: Yeah, at the end of the meeting, Hijis volunteered to modify the dapper to include all of this.

Adolfo Brandes: I haven't checked if that's already done and if it's done exactly according to what we've decided. Right. So the Yeah.

Feanil Patel: It's not Yeah,…

Feanil Patel: it hasn't been updated on the deer yet, which is why I wanted to make sure we talked about it because that deer will get updated in the next day or two. It's on Reggie's display to do so. there's a bunch of other useful information in that deer already. So, the deer is useful to review. I think the deer currently as stated is more aggressive than what the plan is by a little bit.

Adolfo Brandes: I saw other hands up like Roberts. I don't know if Okay.

Robert Raposa: My first question was about the deer, but then it started going on and here's the deer ticket.

Feanil Patel: 

Feanil Patel: This was announced and…

Robert Raposa: But just to be clear, I think in terms of deer process, we're at proposal stage, So everything that you just described is a proposed plan. it's not this or did this with that schedule?

Feanil Patel: accepted. Yeah.

Robert Raposa: I just didn't remember.

Feanil Patel: It was announced and accepted with a more aggressive schedule and we are updating it with a less aggressive schedule. So, I don't think we need to It's going to get reannounced, but it's not like going to be delayed.

Robert Raposa: Got it. I just didn't remember part of it, but I wasn't paying close.

Jeremy Ristau: 

Jeremy Ristau: There is a proposed timeline section of the deer…

Feanil Patel: Yeah. Yeah. Yeah.

Jeremy Ristau: which tries to outline this part.

Robert Raposa: And then I'll try to add notes to the deer as well. this is just an inform basically of because I was once involved in a several week potentially month performance improvements required to actually make forums not fall down on http://edex.org . just ensuring that there is the proper level of load testing of this to that level which is a very large level and…

Feanil Patel: 

Feanil Patel: Yeah. Yeah.

Robert Raposa: maybe that's already done and maybe that's not anyway

Feanil Patel: Yeah. for reference, not to proxy a human for quality, but Dave Worms has been sort of is reviewing and shephering this through. so I'm fairly confident he's thought about performance in this space, but that doesn't mean you shouldn't ask about it. It just hopefully helps alleviate some of the concerns.

Robert Raposa: Yeah, agree. and…

Jeremy Ristau: Yeah, I think that's

Robert Raposa: I'm certain he has thought about it. And there's a question between thinking about it and actually running it through at that level before we turn it on in http://edex.org and…

Adolfo Brandes: 

Feanil Patel: Yeah. Yeah. Yeah. Yeah.

Robert Raposa: oops, you can't go back from mongo to my myo and…

Feanil Patel: Yeah. Yeah.

Robert Raposa: you're just

Adolfo Brandes: There's a little bit.

Feanil Patel: There's two sort of Yeah.

Adolfo Brandes: Yeah, go ahead, Fel.

Feanil Patel: Yeah. there's two sort of major waffle flags that will help not automatically make this go out for http://edex.org when the code lands. I know it's on his plate to sort of communicate and probably for you guys to set those before that code rolls out.

Jeremy Ristau: 

Jeremy Ristau: We have Yep.

Feanil Patel: Okay.

Jeremy Ristau: I think what Robert is saying and what I might request is can there be acceptance criteria inside of this work that is actually run load testing and…

Jeremy Ristau: produce results and have those available to confirm that it does handle a certain amount of load. and…

Feanil Patel: Okay. Yeah.

Jeremy Ristau: to quote Robert I'm sure he's thought about it but there's a difference between trying to prevent and then proving that you can't handle it. Yep.

Feanil Patel: Yeah, that's fair. I think those are great questions to put on the deer and for Ries and Dave to respond to. I know that Edley has been running this for I think even a couple of their production instances already. So I don't know what the size and shape of that is but I know it's sort of like from a trust perspective that's another data point that might be useful.

00:20:00

Feanil Patel: 

Feanil Patel: Cool. Yeah, I just wanted to mention that because I know that there's the sort of initial impact of this real code landing, but then there is the six months from now being able to use the non Python code to talk to the Mongo forums the ability to read from Mongo at all will go away in six months. So migrating the data and oper making sure that that's a thing that's on your radar for the next six months is the big thing that I wanted to cover.

Adolfo Brandes: 

Feanil Patel: But with the course waffles, hopefully you guys can roll it out slowly and make sure that it does the things you need it to do. And the team that worked on it is on standby for bug fixing. they're dedicated engineers. Yeah.

Adolfo Brandes: Yeah, that was a big point from the meeting this is all new.

Adolfo Brandes: We're all seasoned engineers. We have no illusions that everything's perfect from day one. and the team are he called it lowkey standby or something meaning there are one or two engineers that are going to be paying attention to this.

Feanil Patel: and we're going to be paying attention to this,…

Adolfo Brandes: Yeah. Yeah.

Feanil Patel: but they are not pageable if you have a paging worthy issue.

Adolfo Brandes: Yeah. It's not like Yeah.

Feanil Patel: I think is what he meant by low key.

Adolfo Brandes: It's not first level support. Yeah.

Feanil Patel: Yeah. But we will have people who are expert in the code to help with bug fixing. and hopefully with the waffle flag to turn on the Python code and turn it off, it'll be easy to roll out and roll back if you need to. It's the switching from Mongo to my SQL that's the trickier part. since it's a way change. Those are all the things I can think of. We're over a little under time, but does anybody else have any general maintenance things to do to talk about? All right.

Feanil Patel: 

Feanil Patel: I guess the forum thing is kind of a little bit since it is edexplatform specific. But are there any edex platform specific things we want to cover other than the two we already covered? Yeah. You guys can have a half hour back. All right,…

Jeremy Ristau: Have a great day everybody.

Feanil Patel: you guys later. Thank you. Bye.

Meeting ended after 00:23:45 👋

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