DASH Behavioral Model WG, 7 Sep 2023

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: DASH Behavioral Model WG Sept 7 2023

Description

Kamil has made changes to SAI Repo, with (for example) build times cut in 1/2 again
SAI Thrift server and tests
Volunteered to separate SAI Meta from libsai

[sonic-net/DASH] Sai attributes for ACL rules should include list of IP prefixes for the Bmv2 compile (Issue #360)

ACL Tags: Support for tags on the control plane
Kristina to clarify this in the HLD
What is it that defines what we should (at libsai) down, vs NB on the API
State this this doesn't trickle down to libsai, reference Orchestrator

Discussed the Issue from Intel checking out the P4 DPDK and checking for an update from Usha. Also discussed the pipeline and behavioral model, the code work done, and the known issues and gaps.

Paul was introduced to the project.

A

As a separate library and so he's going to be fleeting that on his uh to-do list, next.

B

C

Originally Vincent I wasn't going to record I thought it would be a shorter meeting today, but um I guess we have a little bit to talk about. We also talked about the fact that um the Usha from Intel are would go ahead and check out the the p40pk and get an update to us, um yeah and then Paul since you're new. Maybe I wanted to show you this.

C

Let me pull it over here.

C

So what what we're talking about here is in in Pipeline and behavioral model um here, there's the PSI folder and there's the bmv2 folder and so bmv2 actually generates uh the sci for us as I hope.

C

I have this right guys and so there's the code work, that's done there and then you'll get the PSI, and the here is the readme part where they talk about the the P4 model and here's um pretty much our known issues and gaps uh area too, and so basically sometimes the tests that are written aren't always able to run in bmv2. So we have a few.

C

If you go to the issues in PR section here, you know we can always look at those, but um I think I'm summarizing that correctly right, Chris, that the gaps, what we've been looking at.

A

C

Okay, yeah so, and and so it's been written up pretty well here,.

C

So I'll go around the room that Marley did. You have anything for today.

A

No, no I'm I'm, good, okay,.

C

I feel like we had a pretty big meeting yesterday. um Did you have anything.

B

uh No I'm good.

C

Okay- and we talked a lot, but yesterday, I opened those two PR's um for um I mean not that it has to do with bmv2, but just letting you know, I open those two PRS about what we talked about yesterday for insertion of content into the um documentation yep thanks for that yeah, yeah, no problem and then uh Marian shy. Did you have anything today.

C

Nope, no, okay, no all right- and we talked about yesterday, um even though we're removing the accol tagging from the P4 code, which was great, we may or may not keep it in the documentation I'm. Looking at that so um yeah. So that's where, where we can land for today,.

A

Are left in the documentation, uh relevance to Sonic did that because uh Sonic still has still will have uh support for tags from the control plan.

C

Okey dokey: let's make a note to myself yeah.

B

That's that's part of the problem. It's not a problem, but it's confusing in the documentation, like the hrd doc I believe, is the upper layer way above lib's eye, so we describe what somewhat we should be doing, but sometimes it's not like I think tag might be an example uh from an outsider. It's it's always been confusing to me black.

B

What defines clearly what we should be doing, how the leaps are down versus what gets done on the Northbound API.

B

Maybe it's just me I, don't know.

C

I can try to fix it. um Vincent and I can I can do PRS or suggestions and have you review them and see see if it makes it more helpful right.

B

I mean I, guess the ACL tag in the documentation can stay as long as we stay somewhere that this is. This doesn't trickle down as opposed to lips eye I. Guess right.

A

B

It's not even optional, it's just the orchestrator I guess expands or does whatever it needs to do right.

C

C

Got it okay great?

C

So if thanks Marion for the feedback there um did you guys have anything else for today.

C

The reason I'm rushing a little bit is we're right in the middle of semester planning as well. So it seems like we're all triple booked this week, but I did want to touch base with you guys. So um if there's nothing else, I can. Let you have time back. Is that, okay, with you guys.

A

C

Yeah yeah, okay, cool and uh yeah, so we'll look for uh um the update from Usha. If you can get it and I'll see you guys later.

C

Thank you. Christian I'll stop Brooklyn! All right! Thanks, guys have a good day.

A

Happy planning, oh.

C