GitLab Kubernetes Migration Working Group (FKA Scaling, Multi-large), 23 Nov 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 2020-11-23 Multi-Large Working Group

Description

No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).

A

Anyways, we checked the account team and they expressed uh some concern on proceeding with gcp.

A

This is because some of the effort that one of our other partners put into um bringing us into the deal, uh as well as some uh free credits, uh so uh we are checking to see if we should go ahead with aws um and check with uh check with jerry and he's open to working aws and next up is. There is checking with brent and steve for their input.

A

You can see they'll list you linked there to the discussion uh vlad you want to go next.

B

Yeah sure uh so, on the pages side, we've completed the migration poc and it looks much simpler than I expected it to be, so I hope to get merged in different mars uh within like this week and next week, and then we just need someone from the sri team to just run the rake task in production.

B

It probably will take a long time, but we want to see how fast it will go and maybe we will add a parallel version and rerun it later.

B

That's it uh and you have another point: next manager.

A

Oh yeah, thanks transitioning to what's happening next uh andrew and I are engaging with support to determine basically the support model, what types of tickets and requests get routed to whom mainly infra or support um so we're getting the discussion started now and we'll we'll keep that going to kind of mainly figure out. You know who owns what going forward for this model um and then make sure we trap it and collect any concerns that arise from from that discussion next up from there.

A

The next item is we're making progress on the distribution team on proving out the operator. This is for the cloud native sort of next-gen version of multi-large. If you will, we are focused on proving out using the helm sdk for the core reconcile loop within the operator. This is the chunk of functionality that essentially looks at what's deployed in kubernetes and then generates any objects or updates to those objects.

A

If they exist to basically make sure it matches the operator's expectations, so this has not necessarily been done before as far as we know, but we're hoping it works well.

A

If this does work, we will then be fanning out across the distribution team to uh build up the rest of the charts, we're starting with just one to make it easy, and this is likely going to be the primary chunk of work uh for the initial data release for the operator, because at this point in time, once we have all the charts in the replication loop, it will be able to deploy and ensure the objects are what it expects, um jason, if I'm speaking, let me know, but I'll keep going otherwise.

A

Also another update we have the pages container, mr, is up and actually in review at this point in time. That's very exciting. Thanks everyone for the hard work there. It's true cross team collaboration. So that's super awesome. Once we have the container, we can then start working on the helm, chart updates for the pages service, and then we can keep making progress.

A

Jason, you, the next item.

C

Vmr is up and in place for having the ability to split traffic at the ingresses. uh I'm waiting for review for this. At this point, um I do apologize in the delay. For that one, it turns out. I had an infinite loophole. That was a pain to track down. uh That's what I get for using a language. That's not really a programming language.

C

um The next thing is that the unified structured logging for the cloud native installations is actively being worked on starting this week as a part of 13.7.

C

Our intent is to start with easy containers and go to any of the more complex ones.

C

If no comments or questions, then I'll give it back to you josh, as you have the only discussion item.

A

Yeah, this was nothing super interesting.

C

A

An excellent item yesterday or last week to create a blueprint for the potential item of uh shipping, mineo and deprecating any kind of local storage I looked around. I couldn't figure out how to add them um so camille. If you want to voice over your comments, that'd be great.

D

um Like this process is pretty like well documented, it actually has the full page in the handbook that describes the process itself, but our an actual blog points are part of the gitlab repo, because this is something that is really affecting mostly gita repository.

D

So what you're really looking for is creating a mr using probably like something of existing, like maybe another blueprint as a as a as a draft, creating a mark with this single magnum file, describing the uh like the high-level concept purpose and maybe indicating like individuals for various aspects of the blueprint, because we define different individuals that can provide with the technical guidance like that, our facilitator of the blueprint and that can provide like, like the dri, uh so take a look at this handbook entry.

D

It's gonna really like give you like very comprehensive information like what to do next. uh If you just if you have like any questions, also, there is like the architecture channel and we additionally have the working group uh architecture workflow.

D

That is also looking over these blueprints and and trying to optimize this process of the architectural groupings. If you would notice something that maybe is suboptimal or engineer, some general improvement that you may have.

A

Okay, awesome thanks I'll go through these and get an mr open. I appreciate the help here.

A

Okay, uh something else we can probably wrap up thanks. Everyone appreciate it have a great week and talk to you next week.

A

So we let's just get started. I added a couple items to the agenda here. First, one is uh andrew thomas and I discussed with the city count team, uh and actually this is this public. I can't remember okay, uh I'll I'll, try and check. I don't think, I'm not sure it.

B

A

B

Safe to assume this is public or.

A

Will be more probably yeah I'll I'll I'll, try it. uh I would be paying someone to take that down. um I'm sorry anyways, we checked the account team and they expressed uh some concern on proceeding with gcp.

A

This is because some of the effort that one of our other partners put into um bringing us into the deal, as well as some uh free credits, uh so uh we are checking to see if we should go ahead with aws um and check with chuck with jerry and he's open to working aws and next up is. There is checking with brent and steve for their input.

A

You can see the issue linked there to the discussion uh vlad you want to go next.

B

Yeah sure uh so, on the pages side, we've completed the migration poc and it looks much simpler than I expected it to be, uh so I hope to get merged in different mars uh within like this week and next week, and then we just need someone from the sri team to just run the rake task in production.

B

It probably will take a long time, but we want to see how fast it will go and maybe we will add a parallel version and rerun it later.

B

That's it uh and you have another point: next manager.

A

Oh yeah, thanks transitioning to what's happening next uh andrew and I are engaging with support to determine basically the support model, what types of tickets and requests get routed to whom mainly infra or support um so we're getting the discussion started now and we'll we'll keep that going to kind of mainly figure out. You know who owns what going forward for this model and then make sure we trap and collect any concerns that arise from from that discussion next up from there.

A

The next item is we're making progress on the distribution team on proving out the operator. This is for the cloud native sort of next-gen version of multi-large. If you will, we are focused on proving out using the helm sdk for the core reconcile loop within the operator. This is the chunk of functionality that uh essentially looks at what's deployed in kubernetes and then generates any objects or updates to those objects.

A

If they exist to basically make sure it matches the operation expectations, so this isn't not necessarily been done before as far as we know, but we're hoping it works well.

A

If this does work, we will then be fanning out across the distribution team to build up the rest of the charts, we're starting with just one to make it easy, and this is likely going to be the primary chunk of work for the initial data release for the operator, because at this point in time, once we have all the charts in the replication loop, it will be able to deploy and ensure the objects are what it expects, um jason, if I'm speaking, let me know, but I'll keep going otherwise.

A

Also another update we have the pages container, mr, is up and actually in review at this point in time. That's very exciting. Thanks everyone for the hard work there. It's true cross team collaboration. So that's super awesome. Once we have the container, we can then start working on the helm, chart updates for the pages service, um and then we can keep making progress.

A

Jason, you, the next item.

C

The mr is up and in place for having the ability to split traffic at the ingresses, uh I'm waiting for review for this. It's at this point. um I do apologize in the delay for that one, it turns out. I had an infinite loophole. That was a pain to track down.

C

That's what I get for using a language. That's not really a programming language.

C

The next thing is that the unified structured logging for the cloud native installations is actively being worked on starting this week as a part of 13.7 intent is to start with easy containers and go to any of the more complex ones.

C

If no comments or questions, then I'll give it back to you josh, as you have the only discussion item.

A

Yeah, this was uh nothing super interesting.

C

A

An excellent item yesterday or last week to create a blueprint for the potential item of uh shipping, mineo and dupricating any kind of local storage I looked around. I couldn't figure out how to add them um so camille. If you want to voice over your comments, that'd be great.

D

um Like this process is pretty like well documented, uh it actually has the full page in the handbook that describes the process itself, but our an actual blog points are part of the gitlab repo, because this is something that is really affecting mostly guitar repository.

D

So what you're really looking for is creating a mr using probably like something of existing, like maybe another blueprint as a as a as a draft, um creating a mark with this single magnum file, describing the uh like the high-level concept purpose and maybe indicating like individuals for various aspects of the blueprint, because we define different individuals that can provide with the technical guidance like that, our facilitator of the blueprint and that can provide like, uh like the dri, uh so uh take a look at this handbook entry.

D

It's gonna really like give you like very comprehensive information like what to do next. uh If you just if you have like any questions, also, there is like the architecture channel and we additionally have the working group uh architecture workflow.

D

That is also looking over these blueprints and and trying to optimize this process of the architectural groupings. If you would notice something that maybe is some optimal or in general, some general improvement that you may have.

A

Okay, awesome thanks I'll go through these and get an mr open. I appreciate the help here.

A

Okay, uh something else you can probably wrap up thanks. Everyone appreciate it have a great week and talk to you next.

A