Open Telemetry Uncategorized, 2 Dec 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 2022-12-02 meeting

Description

Open Telemetry's Personal Meeting Room

A

A

I have to say I'm a little afraid of saying hello again last time: I did it it disconnected. You.

B

A

Dot is an unexpected feature of Zoom.

A

Hash account, can you are you able to unmute yourself? Tyler cannot.

C

A

I was away removed, oh yeah, no, no worries you're, able to unmute yourself great. At least you can because Tyler can't. Oh.

A

Hell- or this may be a a reboot- your computer situation I'm not not really sure how to handle this.

A

How you doing chicken it's been a while.

C

Yeah, are you doing.

A

Yeah yeah good good, no Tyler's, trying once again to rejoin let's see if he can unmute this time.

A

So we've lost with last Jack for today, because his power went out, Tyler is muted permanently.

A

This will be. This will be a very uh a very strange meeting. I guess we'll just we'll just.

B

C

B

A

Tyler I, don't I, don't know how you become the host.

A

All right, Tristan also cannot make it today. Well.

A

This is not going well.

B

Come on I, don't.

A

Know if I have the power to change anything in this meeting, I feel like I. Don't.

C

um What is I I could like? Why is he having issue.

A

I yeah I have no idea uh he's getting an error message: it's in the hotel, config file uh chat room about hosts, not allowing participants to unmute themselves, and you can't do anything about it.

A

Yeah, it's kind of uh kind of weird uh show manage participants.

A

Yeah, it's weird because it looks like Tyler is the host of the meeting.

A

Oh you just clean toast.

A

Still having the error.

A

All right, we have two Tyler's now.

A

D

A

D

Okay, yeah I'm on my phone.

D

Can be a problem I guess, um let me see I wonder if I can even share my screen.

A

You know it's okay, I can I can do the screen share. Okay,.

D

A

All good there's only there's only a couple of items to talk about, um but I I do yeah. Okay, let me let me do screen share here and then we can kind of get started.

B

A

um Yeah so I guess the the first item conveniently since three out of the five original people in this group aren't in this meeting it's probably timely to talk about rescheduling the meeting, um so I put together a a poll on the um in the channel.

A

um The two dates that got the most votes are either Mondays at 8, A.M or Tuesdays at 9am. um This is all in Pacific time.

A

I will say that I think we should probably go with Monday um just because that actually included four of the five original contributors to this working group, uh whereas the Tuesday only had Tristan and I so um I guess. Unless people have a problem with Monday at 8am, PT I would suggest we would go with this. One yeah.

D

I mean that sounds good. To me. um I mean it's not ideal, obviously, but just it.

A

Is what it is right right? It's like there's, no there's no good time with all of the other meetings that are in hotel and all the other working groups and stuff that's happening so yeah.

A

um Sure cons is that a time that you can you can make I guess it would be like eight eight p.m. Ish your time, if I'm not mistaken,.

C

Yeah it works for me. um Maybe.

B

C

Time around, like seven to ten it usually it's the same time eight pm or something around here. It works for me. Okay,.

A

Awesome I mean it'll, be better than uh 10 p.m. Fridays so.

A

Cool so we'll I'll take an action item to update the community issue with it.

B

A

um The other thing that I wanted to just make a quick decision on I think we just need to make a decision on the casing, for the configuration that we want to use. There's a bunch of PRS that either have like a dash in them or an underscore.

A

We could also use camel casing: I, don't really care I just really want us to come up with a decision so that we can move forward with it and not have all of these uh mixed casings uh being proposed. So.

D

Yeah I I, agree: I, think that uh the pr to the spec um can have the bike show discussion over. There needs to be cam, okay, snake case or you know whatever you want, but um I would uh I would say Alex. You should take the authority here and just decide and tell everyone what we're gonna do and then put the caveat or the uh this asterisk. There saying like this is up for discussion. You know at this stage in the development or something like that.

D

All right yeah, because I I mean the thing that I I, think that everyone can agree on is consistency like that's the thing that everyone wants. um I mean I've written in so many different languages that it's not really an issue to for me to switch. The naming I, just like I agree like just one is, is great and if we need to change that later on, like I have no problem going through and renaming things like.

A

All right, I'm going to propose, we just go with camel casing to be consistent with kubernetes and perfect. There. We go all right since there's only three of us. If two of us can vote towards us, then we're good.

A

All right there, quick decisions, yep done uh if people want to argue about it later, we can, like you said you can just bike shut in the pr for the spec, because that's likely to happen anyway. So there's no point in having it here as well. Right.

D

A

The other, the last thing I wanted to bring up so there's a proposal here.

A

um I think this was based on some of the comments that strikant you had made around um how we can reuse some of the exporters.

A

um If, if folks have a moment to review this, PR I would like to kind of move this forward, and we can then um actually start using the uh the config yaml to apply some of the schema testing that we wanted to do.

D

um This sounds good. Are we? Are we able to validate that? The exporter field is an existing exporter so like in this example, we have zip and listed here like I know in The Collector, it's done via like code. Is that something you can do in the schema.

A

um That is a good question. um I would throw that in this issue here. I suspect that it's something we might have to do through custom uh validation um but I'm, not I'm, not 100, sure I, don't know what the extent of what Q supports is.

D

Yeah I think I think uh that's a good question actually yeah, because there's a lot of like um procedural ordering that could be confusing there. uh I, don't know I. Think Json schema um might have a tough time with this, um but I'd have to have to double check.

A

Right, I, don't I, don't imagine that protobufs would support this out of the box um either. You'd have to just specify a custom martial or that's what we do in The Collector today right so right.

D

Okay, I mean I'm not opposed to this style um again. Also, not even at this point, I don't know if it's like critical that it's structured this way, um but yeah I think it's something to admit. I'll, ask I'm typing of a question.

A

Okay, yeah I suspect that this like at this stage, we just need to have some stuff that we move forward on, um and then you know if it turns out that it's such a pain that we would have to write a ton of custom code in every language. Then you know we can just put that as one of the known issues of things of this particular schema has and kind of move forward with it.

D

I agree yeah. That sounds good to me.

A

C

um I had one thing to um add like wanted to ask here so there's an exporter argument in the schema right for the batch processor. I was wondering like if you could make it to like make it a exporters like uh and allow multiple copy uh configured for single batch processor, like the hotel underscore Trace is exporter. It currently supports the commas. Operated values. Right um I was wondering if, if it's something that we want to consider uh enabling here as well.

A

Right, I guess the so today that variable um implicitly expect someone to configure multiple span processors right.

C

um Yes, that's true like uh so the sdks behave in a way that for each of the exporters configured, um it uses the same processor. In this case it can, it can be a batch processor uh with let's say, Japan and otlb. So if you, if you, if you configure like the exporter sections to to be included like multiple the same cluster will be used.

C

If you are, if you want to um add a new like different processor or a processor with different configuration, um adding a new section is fine, but if, if it's just the same, configuration I think duplicating the whole processor, config again for another exporter um doesn't seem like a nice experience.

D

Sure, I don't think, that's necessarily something we need to solve here. I think duplicating.

C

D

Might be fun for this part of the specifications is we're talking about the schema and we need to validate this I I. Think if there's something, if you are going to say like something like yaml or Json, you can always use meta programming here and use. Some sort of uh you know, stored procedure to generate the the reused element as well, so, like I, think there's details there that don't really have to get solitary, I, think export or how we have it in.

D

This is just fine like again, because it's it's we're trying to validate if a schema can validate this. This configuration. Not necessarily is this the ideal configuration.

C

Foreign, that's a good point. I think.

A

C

That can be changed, but it does not have to be something that needs to be addressed in this.

A

Okay, um yeah I guess otherwise, there's not been a ton of progress here. um I haven't seen anything around the q line, schema uh validation and I. Don't know Tyler if you've talked to Tristan at all about any any progress there. I did I did merge this PR into the main branch just so that we can kind of Move Along um so that people could see that there is a q-line folder in here, but um yeah I'm. Not sure do you know if there's any any progress there.

D

I know he's been working on. It uh he's, regrettably, always working on it at the end of the week similar to me, and so it's like right, it keeps getting pushed back, uh but yeah he says he's got some progress. I talked with him just on slack just before the meeting um he's looking to find some more time to knock it out uh completely later today. But you know that probably means that early next week, uh based on you know, engineering timelines, um so I I imagine yeah he's working on it.

D

I also I mean I'm just coming off of a lot of vacation, but um I've been looking back at the original doc um for the uh eventual Otep and I I still need to that's. That's on my plate of things. I want to keep updating, so yeah he's, definitely working on it um and I'm I'm not actively but I'm, trying to get to working on the the Otep doc as well. So um just kind of a status update for people that aren't doing things.

D

B

B

um That being said, like I, do.

D

Know he's been looking at it a lot I think more just in like the the extent of cue um but yeah he's I think trying to get something pushed out. I did tell him that you merged at PR and he was excited because it's like you said like incremental, and somebody can see it. You know.

A

Yeah we just kind of have to keep keep moving forward like we'll, have plenty of time to get bogged down and, like I said bike shedding and other like. We just need to propose something, and then we can kind of iterate over it. um There is there is this one issue that um pallorad opened around uh versioning, so I I, foolishly thought that merging the version uh attribute would be on uh would be easy and um not up for a lot of debate.

A

um So I just proposed this as like version- and you know some uh patch, minor and major versions, there is some questions of whether or not we need the patch number at all. There was some questions of whether or not we need a minor version. I think we should keep minor and major version just because then it allows us to make changes to the configuration that are backwards compatible. So if we were to add new functionality or whatever, then we could just bump the minor version without having to bump the major version.

A

There's also a question of whether this should be called scheme: Dash version instead of just version, because version may be confusing to someone who thinks that this is a configuration file version um I'm, I'm, okay, with calling it I suggest a schema version instead of scheme version. But that's just my preference I think schema is more meaningful here.

A

um So I guess. If anybody wants to add a their comments on this issue, just so that we can create a PR and move forward, um whether we, whether we need to keep only major or major, minor or major minor patch, if people have a good use case for that, but.

C

um I'm not sure if uh the schema I think schema version will be confusing, because there's already a schema name, schema version Concept in the specification for the the semantic conversion schema that that's got published.

C

um I think it will be confusing like there's there's this here, the schema version and one in specification, I I think it will be confusing um right.

A

Right because I guess, then we would be adding schema URLs uh in this configuration file for resources, for example, and then what does that mean? As far as the like? What what schema are we talking about.

C

D

Yeah I, don't know I I, don't see. This is really that important honestly um for the Otep I think that people are getting really focused on the configuration file and what it's going to look like um I mean I. Think yeah, like all these details, I think, are things that are great to have a discussion on, but I I just needed. You know test to say like we're going to build a configuration file defined by this schema in the specification and then when we build that specific error that configuration um file in the specification.

D

This is a great topic to have a conversation on I. Just want to make sure that the schema language we're using to Define that in a specification, can handle this stuff, which I like I. Don't.

C

D

Really much of a discussion here around whether to handle it like I think it can like I, don't see too much of an issue there.

B

D

Which I think is uh yeah I I think that I really need to get after that doc, just to get a um more clarity on what our Otep is going to look like, rather than um I, think we're having great discussions here. But it's your point, Alex I think we're gonna just have the same discussion when the broader Community comes involved in this, so um yeah. Okay, all right I'll have to dig into this a little bit more.

A

Yeah I mean, if you want, if you wanted to I, was I, was actually considering, uh starting to put together like a markdown file in this repo that would be kind of the source for the Otep. um That way, we can just start writing it out here and kind of scoping. What what it is that this hotel will contain yeah.

D

That's a great idea.

A

Yeah, so um maybe I can I. I was already thinking about this in the context of wanting to identify the limitations of what this configuration uh will support and won't support.

A

So, for example, like multiple Tracer providers is like a no limitation of something we don't want to support in the in the config out of the box, so um yeah so I can I can kind of start uh like just committing like a template with some very little bit of information and maybe pulling some of that from the working Dock, and so we can kind of start having some progress here.

D

Yeah I agree, I think if the the Google Doc was a great place to to start and get a brain dump but uh yeah as we structure this, let's have some consensus as to what the Otep is I think it would also help Focus the development effort onto what this you know is hoping to produce and that's the the Otep so I 100.

D

If, if you can get something in I, will, if you don't I'll, try to I'm trying to work on this this afternoon, um I'd have a limited amount of time, but uh definitely next week is something I'm trying to focus more on this um by early next week.

A

Okay, yeah, you know what I'll I'll probably do it right after this meeting uh just so, we can have at least a starting point and yeah. Let's just get it merged quickly, so just so we're not like yeah yeah. We.

D

Don't need to go through it. I will definitely be looking for your PR, then, and let's try to get this merged today as well, but.

A

Okay, that sounds good.

B

A

There are no other items in the agenda. I see, there's a couple more people that that have joined any any questions or comments and thoughts that people have that they want to share before we call this meeting.

C

You know not me.

A

Okay, I guess we can. We can call it and get back to work. Then. Okay, thanks everyone all right.

C

A