Open Telemetry Uncategorized, 8 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 2022-11-08 meeting

Description

Instrumentation: Messaging

A

Enough morning,.

B

A

It's early excuse.

B

B

Nice and chilly here today too, just below freezing last night. So oh.

A

Yeah yeah I think it's supposed to be freezing here tomorrow, night.

B

Yeah I think next Monday is now the first night. It's not going to get below freezing. So oh.

A

Wow yeah, but we've had like we've had like uh three weeks of rain, non-stop, pretty much yeah and today and today. Finally, it seems like no rain today. So that's good yep.

B

A

A

B

Think Santa say it wasn't gonna get here.

B

I, don't see anything from RAM on my side.

A

Looks like around has a couple of things from the agenda so.

B

B

um Priced up Ryan my work program uh they're, mostly just I- have some PRS out. So the first one is the event.data that we talked about last week. Yeah.

A

Yeah I'm, looking at it right now, yeah yeah.

B

um So I I reviewed the cloud events which came up in the spec meeting in the previous week um and it's it's very closely aligned to what we're doing with events uh so I pinched some of their wording uh for that they also defined and a pick a data schema um which we've sort of talked about a little bit and originally I, had it in this PR and then I pulled it out. Thinking uh the New York in Telemetry world.

B

It's always fun to try and cut things back rather than trying to put in too much, um but that's one additional concept that I thought we could have, which we have talked about before is saying for this domain name combination. You can also Define a schema and that schema can then be used to you know effectively provide validation or some context to the back end. So it can do stuff.

B

um I was going to bring up the log seek tomorrow to get uh tigrid's view to see whether he wants to try and drag that in. At the same time, uh in the spec meeting this morning, uh there's someone trying to effectively add a payload attribute uh to a span which is very similar to what this would be so I sort of linked it in this in the spec, as well as the pr itself. Sorry.

A

B

um Yeah they wanted to create a effectively uh semantic conventions for attribute payload attributes, um and it's part of that they need nested attributes support okay, um but yeah. It wasn't going down very well. Ted was in the meeting and was driving when I I were absolutely late, but um that's an interesting thing.

B

I I proposed this is a solution where we say they could create a payload events and then use the event data as a um a mechanism for what they're currently displaying payload attribute, which is a blob of data which makes sense in terms of licking in where we're going.

A

So for this schema are you thinking introducing new attribute named attribute like event, dot schema or something like that or.

B

Yep yeah, a red dot schema or if we stick with the cloud event today, if you have data schema, so we could say event.data schema and that way we say the data schema of a cloud event is the same. That's what we have here, I.

A

B

um But yeah I I removed that out of that PR, because I thought was going to generate more of a discussion than just getting the data object in um I. See Santos just had a quick look at it, um but I don't know if he has any comments. I've marked it as draft for now. The other issue that's coming up in that PR is they're now validating the yaml files, and even if I don't modify the yaml files, um there is no way in the schema to Define nested attributes.

B

So therefore, I can't add event.data and the existing ones, for both the exception and events are complaining that they have no type, because there's only a small subset of types there has to be a span and resource Etc, so um that will probably have to be resolved at some point as well.

B

Okay, um that's the first one. The second one is just I finally got the sandbox mode script working so that PR is effectively just to fix the script. um I have actually already used that script. If you look at the history of the commits, I did some I ran. The script did some manual merging back around the script a couple of times and merge them together.

B

um So you'll see the output perfectly all the stuff that says, Auto merge was generated by this script.

A

B

So I've already fixed up the existing history because it was a lot worse than I thought. I tried manually, fixing it um and then, when I got working with this script yeah there was quite a few that were just wrong um and that's because of the way the country, repo and get merge works.

B

um I did have an earlier version, but every time, even if it did nothing, it would generate six commits. So I brought this one back. So this one now generates one commit um or only work, one or two commits depending on what it's doing. So all my comments are there are to try and just give you context of what. Why and how so, okay.

A

Perfect yeah I'll take a look at it.

B

Cool uh Hey, Ram, I, guess you're next.

C

Cool um hey guys morning, um sorry I came in a little late now, but uh listen to most of the conversation uh there and there I did put a couple of comments there in a no big deal. But um you know just take a look when you, when you can yeah I, think it's a good good start.

C

um Ned probably mentioned you know we had I know never and I. Had a discussion on Thursday talked about how you know this needs to go in before we can um at least be out there. So we can reference uh ewrs, uh so I'm, working on um creating another PR for the already defined schemas in the semantic conventions, um spec repo.

C

um It's should have um dropped by end of week um this week. um So that's basically the update there um and the next one is um I've been talking about uh recruiting supports from internal. uh You know, teams to come and participate in the schema definitions and stuff. We did have a good meeting uh last week um and um a team called hard analytics data, scientists and stuff. You know: they've signed up a couple of their analysts.

C

um They're still working through. You know who and plan commitment and things, but I expect them to show up um in the next few weeks. I'll, you know we'll keep following up and pushing him um hopefully be helpful for us.

C

That's basically it um I um I, don't know if uh you know we want to continue the schema uh for other events and stuff or the product for tomorrow. uh Up to you guys.

A

C

Think I'm, okay,.

A

With referring it to tomorrow,.

B

C

A

C

Okay sounds good I.

A

Have a couple questions for your Rams so, like the the pr that you're talking about uh that you want to open this week, is that is that for the structure of the um the semantic conventions for events, correct, okay,.

C

B

Most specific events as well.

C

So I believe we've done um three events right page view: Fetch and exceptions are done. um Let's, let's try and get them specked out there and then reviewed by larger audience.

C

um That'll also give us the opportunity to make sure that we get the finalized names and stuff uh listed there and then also serve hopefully serve as an example for when we push the event.data national attributes uh thing that Nev's putting up that's the that's ID yeah.

A

Okay, so so I imagine I. Imagine there will probably be uh comments about the structure of this, like I can General comments about the structure and then there's gonna be probably people commenting on the specific attributes, correct.

C

A

I wonder like if, um if it would make sense to open them at separate PRS, you know so that, like uh discussion like on one type of event, doesn't block the other ones. um I.

C

See so uh if I heard it correctly um say, let's say we have three events. You know three separate PRS yeah.

B

C

um Or do you also want to you know possibly clear the the folder structure that we talked about like in a browser under you know, it's like I forget what it you know, the the folder structure that we want to create under semantic conventions uh under laws um that is a separate PR, could give people.

C

um You know opportunity to comment on just that structure. Is that what you're thinking Madam.

A

Yeah yeah, okay,.

C

All right, I yeah, we can do placeholders. That's a proposal to start defining events underneath this folder structure and then separate peers to as a follow-up to actually have meet.

B

With the detail, yeah sounds.

C

Good I can try that yeah go ahead and.

A

I had a question about your um analytics team, yep, uh so they're, the team- that's uh that's like ingesting the data and so they're like their intent. Just to like make sure that um the semantics make make sense to them from ingest perspective.

B

They're more the consumers of the data so so effectively our team uh lets it get ingested into the back end services and then they've got the the data scientists and the analysts to go off and create views from that data. If that makes sense, yeah.

A

B

C

B

I I correctly described yeah.

C

Yeah absolutely yeah yeah. Exactly yes, that's great yeah.

A

Okay yeah, so my I guess my question was like: are they going to contribute uh with defining new events going forward or are we going to revisit the the events that we've already discussed.

C

um I I hope, there's there's going to be a lot less. um You know revisiting our existing events. um You know um they.

C

That's my thinking, you know these. These guys don't have um uh because we use uh you know we used our internal events as the basis for this right now. Those are the internal elements that they are considering today and are able to. um You know, satisfy their. um You know dashboarding requirements and analytics and everything. So there shouldn't be a huge gap there.

C

um The the thinking is, they could potentially, you know, uh come up with new events that we should start considering and defining for the community, and they could also use this opportunity to basically say Hey. You know, we've been using, for example, page view that um you know uh um our team right. My my team and its team has defined um they've been using it, but probably um you know, um you know, think.

A

That additional attributes.

C

Might be helpful or whatever never an opportunity now they never had an opportunity to go back and predefine our internal events, but now it's uh it's a great opportunity to Define them uh in open Telemetry, um because it's available in the ground floor right. uh Defining these things. That's.

A

C

Don't believe there will be a lot of these drastic input coming from them. Asking us to kind of go back to the drawing board shouldn't be like anything like that.

A

C

Okay, it's not a completely different thing; uh they are consuming our data, um the schema of which is what we have used as input uh for open Elementary.

B

Okay yeah, so so, like the user action page action stuff, um they were heavily involved in in some of the the names that we end up using uh in the open in the click analytics one. So yeah um I think they have some other things. They would like to do uh at the moment. They're they're, mostly on the the JavaScript side, but by defining an open Telemetry it effectively opens the door for them to support other types of clients, so I think they're interested in doing that and therefore defining helping Define those events I see.

B

C

Yep, um if, if it helps you know um we, you know we are actively trying to get. uh You know. People like that involved in the in the community, so um they probably have um uh there probably will be observing for some time before, jumping in and start going to go to the Olympics. So um the active, active contributions will probably take a little bit of time.

C

Yeah we're not going to wait later on by one thing: you can get a small phone okay, uh so very quick I want to show this structure I believe this is what you're talking about. We discussed briefly uh Martin. uh You know based on Europe, um that was in a draft uh close now. I think uh this is what we are going for. I wanted to give a quick preview for this group here uh under you know, uh logs um semantic Dimension.

C

You know we create a new thing called browsers, and these will be here so I right now. This is what I have, but I could play a placeholder. So this is what you're thinking right. We can put these things in and then push this as a PR see if people have push back or anything like that.

A

C

B

But we can Define that that folder is the the domain. So when people ask you, what's the structure, we can say well, it's in logs the browser folder uh is effectively the same as the domain. Okay,.

C

Yeah sounds good page view. I think I've started putting comma. You know content here, but we can just to a placeholder and push this in. Okay, all right.

A

Great thanks, thanks for doing that and yeah, if there's anything that I can help with just let me know.

C

um Otherwise, I'll.

A

Just wait for the pr and go okay.

C

A

Also, just just so, you know, uh I'll be I'll, be on vacation next week, um so I'm I'm still here this week, but then I won't be back until the following the Thanksgiving week. So.

B

Yeah I I, guess I'm here this week next week and then I'm out for two weeks, um so I won't be back until the 5th of December, uh then I'm here for two weeks and then out for two weeks yeah for a few weeks with.

C

A lot of holidays- and uh you know people came off coming up and things uh by this section. We just you know quickly, put our timeouts into this stuff and then see what we want to do and practically we could cancel some of the meeting good stuff um yeah. Should we do that.

B

uh Yeah I know the maintainers and stuff have a calendar where we want to put out yeah.

C

We can do that um yeah.

B

There is potential security issues, but highlighting when people are not yeah.

C

Okay, I guess you know we can discuss this as a topic say tomorrow or something and then decide when you know which days we're gonna meet.

C

Okay, we don't have to write anything down about uh when uh we're gonna be gone or whatever, but we did uh discuss that tomorrow. I guess.

B

B

Yeah that makes sense.

A

Around just one of the next just quickly quick, let's say uh I, think this is a great progress with these with defining these events, and thank you for leading this. Oh.

C

Yeah I appreciate it. Yeah I think it's a collaborative effort, yeah we're happy to do that. Thanks.

A

Okay, great well, um yeah I, think there's probably nothing else. Let's just postpone the further further.

C

A

Discussion for tomorrow so I think there's nothing else for today.

C

All right thanks.

A

Thank you cheers.