Istio Config Working Group, 12 Apr 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Config Working Group 4.12.18

Description

https://docs.google.com/document/d/1P3p7zOpX66hPoZBi_CiC36JW7JmoaLWqE2sgHvdq5tY/#heading=h.1ch4fhmjmezx

A

A

We have only one patent: hey.

A

How are you fine I named you like, okay, so I think we're gonna have a relatively short meeting today, so just to go quick, go over the work item, tracking, Jason and happy: don't make server-side revolution yeah.

B

B

Okay, once that's approved, then we'll have some server-side validation in place off by default. There's some follow-up ers to clean things up mixture I, take it that too long by default or the point a would be off by default, some try to get it to Naples and then testing at least for a month.

B

A

So these are all bundled together right by this associate change, mixer social change, the queue meditation investigate see are these cute server-side validation? Yes,.

B

That documentation is it okay, at least the core functionality, though, is there okay.

A

Okay, it's for a galley alpha I have a peer out for doing the research thinking. I am going to the code. Reviews and I'll have another follow-up that completes the binary like full implementation. Its same time, I am trying to write an integration test for the whole thing so that we can, at least you know, have some good coverage.

C

Joy is out this week, so I wanted to chat so yeah, so so galley is gonna, be in 1.0 and we're gonna use it as a to support the validation, yes stuff, but not the duty conversion. Before.

C

A

B

And if somebody, because if somebody wants to play around yeah.

D

B

It's not they're, not the computing community, real authority, everybody ages. We can figure that.

A

Sounds good okay, so the other one I think yeah on the other that we can discuss is the the package and see hardly reorganization and mixer Christopher. Do you have any obvious on it?.

A

C

Yeah you too far away you just.

D

C

D

C

Thing it was up.

D

Yeah, we talked about just changing that one piece such that, instead of passing in a string, we pass in a map and that's fairly minimal change. I have that test. I'm testing that locally. Currently it seems to be working so I'll push it up for review, but that doesn't really bias much of anything and actually completing the task. I think still with Mandar.

D

There is changes that need to happen in the pro dose so that we can actually change, naming and consolidate the CR DS, but based on just some back-and-forth emails, it seems like something where he don't need in 0.8. The perhaps push this out so 0.9.

C

Okay, yeah I think we talked about combining that with the.

C

Like removing the per adapter series yeah so.

A

Both work, okay, justice in some context, mixer, has custom configurations. He are these for each type of adapter, I'd instance. The plan for mixer is to actually consolidate them so that there is only one type of instance, CRD and then there's only one type of handler series, so that is a work on under state and Chris is actually trying to make sure that we align the API group names and things like that right. So both are disruptive user.

A

Visible changes, that's gonna, be this gonna have back backwards, compatibility related problems with previous deployments, so it's probably better to just do this, like you know, yank, tube and date, only once and not twice. So we are trying to coordinate among these at each other to actually get that work done. So do.

C

We need a branch for this, maybe your to do both sets of work, I think up.

A

To the point of the actual changes we can like, I can become checking a bunch of a couple of CRS without disrupting things. But after that, yes.

C

Okay, okay, so either we put it somehow put it behind a command line flag or if that's too much work, we have to do it separately and check it in a common place. Yeah.

A

I think so the in an email like in so Victor, if you do the plumbing of API group, to kind mapping yeah that actually gives us a nice way to put things behind the flag. So.

C

Yeah but the other stuff, the commander's work, yeah, that's put that behind the flag. You can.

A

Right so we can actually it's probably it's gonna, be probably card a lot of temporary Torah code just to support that place.

A

All right, so any questions about the work track.

A

A

All right, so these config fields default values so I'm going to the discussion list. Is there any stakeholders for this online I leave Joe and the M would be the main stakeholders but I'm not very familiar with this problem. I just read the bug description.

A

Unless we have the stakeholders I'm gonna, move on to the no other one.

C

But I did have a discussion yesterday with the M and.

C

I guess it's not not quite related, but Lea. They want the same concept. This mixture of global and service stuff for our back support, so I told them to just follow the mixer model, so they wanted to have global resources and local resources in our at least. Let's be consistent. Dedicate.

D

C

Put this stuff in there and that's so they'll they'll use the same thing spaces as mix with those so we'll have that set the expectation that they'll will all get that out in a couple months once we have our our key outside the outside this to you, okay, but for 1.0 it's gonna be consistent at least would make sense.

A

So let me have its outside.

A

C

So on yeah, they just want to avoid they want to be able to that's. One thing we need to consider is: if we have a hierarchy, that's outside of this tier, where we compose these things together, there's gonna be some way for stuff at the top to be forced down and.

D

C

I, dabble and some other stuff can be overwritten, so the security, the security people can just say this is required for the effort entire mesh and the service provider that the bottom can't change. That I think that's the kind of stuff that they're looking for. Okay,.

A

All right so allow individual API group inversion for each country source in mixture. So this is the Christopher. Did you add this one.

D

Most likely we just covered that there so okay.

A

So is there anything further that need to discuss on this, not.

D

A

Are there any discussion, topics, issues that we should talk about.

A

All right, all right, I guess we're done. Okay,.

D

C