OpenCrowbar Crowbar, 5 Sep 2013

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Crowbar Sprint Planning 2013 09 05

Description

Crowbar Project Planning for Sprint 2013-09-05

A

So adam, I guess you can do that anytime.

B

um Yeah sure um I mean if you want, I can do a quick demo now.

A

No, I think we should do the demo next wednesday. I want more people to be able to see it and so on and so forth. So, since it's only robbing me that it would probably not be well.

C

A

Not not from from the dell side, you know it's only the two of us, so I I would like to have other people skate as well.

A

C

A

Is there anything you want to talk about rob, uh there's a couple of things. I have the um agenda list that we started. Did anybody add to it.

A

A

So the demo is on 911.

A

That's what it sounds like: okay,.

A

And so we're moving we're moving this to uh the crowbar.

A

You're right the the move, it will be moved to the promo repo because, um oh well, I think, there's a better space for it. I think adam corrected the copyright issues and those by the way they don't bother me too much um only problem is that. uh Well you know I I mean I think it's your copyright, so it should be added to it, but that's all, um but yeah so he's gonna go and and and move it over into the crowbar repo.

A

Cool the next thing, let's say: first thing: I had to list the status of the pebbles branch and pulls with it, so I can go over and we can. We can re-prioritize the list of items that I had status of pebbles. Let's go ucs sounds like you talked about it, documentation, havana, branch.

A

The mesa branch, a hyper-v pull request- and I would add to this now the um the rel pull requests.

A

uh Status of ci progress, I'm not sure we're ready to talk about that. If we know the right people um and then crowbar to update some status.

A

I was in pro and then um just the shift. If, if we're going to keep this neat, these meetings shifted by week, we should just resolve that.

A

Any any changes or additions.

A

Not from me adam, I don't know, we there's the calls showing that we've got uh a whole bunch of people, but only adam is identified.

B

Oh, no there's a bunch of us from from our side.

A

Okay, sorry guys yeah, there's a holiday uh for some of our team members, so they're they're out judd and arkady specifically.

A

B

I've done that repository transfer.

B

That's it's already.

A

Just like that,.

B

Yeah, it's very easy on github.

A

Does that also mean that the ports get updated as well or no.

B

um Well so so the the forks of my repository are now forks of the crowbar one. So in particular the souza fork is now looks like it's derived from crowbar, just like all the other ones, which is what we wanted.

A

So they update that automatically.

B

Yeah, it just changes the references here.

A

uh So the backlog, the backlog of pulls for uh crowbar, looks pretty reasonable right now, although it's making progress, the hyper-v.

A

There's a couple of big classes of work in this, which is the rel centaurus.

A

Polls that are that are in there.

A

Are you guys, are you guys doing a review on that? I know that we will. How do you? How do you want to bring them in? Do you want to bring them as a group? What are you thinking.

D

So I I did a quick review on these, and uh so there were a few issues and I think they got fixed. But I didn't do a second review after the fixes.

A

Okay, so that we we can just work these through the normal process. We don't need additional discussion.

D

No, I don't think so. I mean they are in general, they are pretty trivial, so.

A

That that was my my thinking on the review too. Since they came in as a bulk, I thought it'd be worth doing discussing it. Hyper-V changes, look like they've been either. I think that most of them came in.

A

You know of any other.

D

The first batch came in, but we have three new put requests for hyper-v. I think now: okay, one for provisioner one for network and one for another yeah.

D

The another one might need some discussion because it's a bit complex, so let me just quickly explain it. um Originally. There was a a new bulk lamp for ipv.

D

That was showing so that it was containing the cookbook for ipv for really running chef and deploying stuff on the windows note, and it was also containing, I would say, the proposal stuff for hyper-v, which was more or less a fork of the the nova bar clamp and had a. I had a discussion with cloud base in octavian and since the nova bar clamp now supports uh assigning nodes to different hypervisor hypervisors.

D

um We decided to just add a new role to the nova bar clamp, and this role would reference the cool books from the hyper-v bar clamp, because we don't want to ship the cool books in the nova buckland because they are really huge, like it contains huge binary files for windows.

A

D

So we want to keep that separate, but uh from our ui perspective it should really belong to the notable clamp.

A

That makes that makes perfect.

D

So there's some: if you look at it, if you review it, there's a small hack at some point where we just check if the ipad cookbook is installed and if it's not installed, we just don't accept using uh the ipad roll.

D

Okay, it's kind of hacky but yeah. If there's any better id on how to implement that, I'm happy to do that.

A

No, I think that makes sense.

A

Cool okay, so that looks like it's it's it's rolling forward. We don't need um additional discussion on it, because it's good.

D

Other than that.

A

Okay, we're trying to spin up a team here that knows hyper v um to do some do some test and review of it against the pebbles branch. Although we'll talk about we'll talk about uh havana work and in a minute, and that this guy will probably want to focus on that now,.

A

Cool this is excellent.

A

Okay, hyper v, so these we've got these covered.

A

uh And just we're we're updating the mesa branch. I know this is more of a stable branch, but there'll be some activity on it, we're updating it to the grizzly three. um We back ported the vlan support that was already in pebbles and then we're going to be doing our hadoop work on off of the mesa branch.

A

So this is a heads up for that, so we're effectively doing patch just some patch releases on that.

A

Cool good um any other topics before I move to havana and branching.

A

Okay, so our expectation is to start work on havana codebase, using pull from source.

A

Potentially as early as today, but I I think it's more likely that it's going to be sometime in the next two weeks um or potentially or even it in two weeks, exactly because that'll be our boundary.

A

There's already there's already work going forward on havana using the pull from source pieces um and the aha work that we've been doing is targeted to be focused towards havana as a havana delivery and the the goal. The goal for this is to be able to have havana.

A

As a community.

A

We know we, this is pretty typical, for one of the objectives uh for us is to be able to deploy the current release at the summit.

A

And so that it supports that. My expectation is that we would do that as a branch off of pebbles. So the goal we need to make sure that pebbles is as stable as possible, because I want to minimize merge issues. Does that make sense? Is that a reasonable approach.

A

Yes, I'm waiting for a response. I think that makes sense yeah I I mean you know, uh and I think it will coincide nicely with um kind of the release of the suitsu now and then uh more of a break fix, pull request scenario rather than a net new.

C

A

Right well, the the consequence would be I'm expecting the pebbles branches to stabilize right. So, let's left new code change, uh most of the new functionality would be focused around pebbles, and so the timing is such that I want to make sure that everything that's in flight, that ideally the biggest changes in pebbles are already in right. Since I know you guys are stabilizing for your 2-0 release, and so structural structural changes should especially the krebs. Our core should be minimal would be my expectation right.

A

Yeah we briefly talked about that before before you were on the call. Actually, so I think okay just locations might help, I'm sorry. What did you? What did you say last time?

A

We briefly talked about about that um before you came on on the call, and I think the expectations you would have in kind of the plans that you were laying out match up with with what what um what adam has said: okay, they're going.

C

To move a little.

A

Slow oil- that's going to basically mostly be you know,.

D

Iterative releases.

A

And pouches and stuff like that to the 2.0 stuff, um it sounds like there isn't for a while a plan to do a major o release and instead move to like a more iterative approach to the things. So I think that should also translate into less frantic activity on the pebbles branch is my guess, and is what somewhat was confirmed by uh adam as well?

A

Okay, because what I would like to be able to do is spend a lot less time. uh You know validating and doing and doing reviews of pebbles.

A

uh We still want to maintain code, quality and stability, because it should become the stable branch. But if there's, if there's still a lot of structural changes, then that's going to be harder to. You know we're going to be splitting effort and that doesn't that doesn't make sense. I just want to make sure we're alive.

E

Just to make sure what kind of structures changes are you thinking about there? I don't think there were any like bigger changes in the last eight weeks, like structural things,.

A

Well, there's there's.

E

There aren't many.

A

And no, it looks like it's been settling, which is why I think it's a it's a good time to do this. um The biggest thing that's been been hanging out. There is the change to add a database to capture sessions so that that poll is still hanging.

A

We should, ideally that is either agreed, not to do it or it's in um before before we fork or before we branch.

A

In that case, and then you know a lot of these, the rel stuff isn't really structural hyper-v was there were structural changes, but I think we've gotten this resolved.

A

um It looks it looks like we're pretty reaching a good time. I want to make sure that what what I'm seeing it matches your expectations.

E

I don't know of any like structure changes. We are planning for the pebbles branch, so I guess we all agree on that.

A

Yeah I mean I I would, I would say even even down to you know any significant ui changes or enhancements, or things like that. We'd wanna, I mean small small stuff. It's not that big a deal, but I I just making sure that we're not we're not creating divergence that we, because what, when we, when we branched mesa, there were a whole bunch of things that have since changed in pebbles and so, which is fine. That's why we why we did the branch it allowed.

C

A

To continue on those on those items.

A

I just want to make sure that we don't.

C

A

Create the same situation where we have leave behind stuff that we want in the in the new branch.

E

A

E

uh Yeah, the probably the biggest change ui wise currently is the open pull request for quantum the cisco support, but other than that, I don't think there is much and I don't think there is much in the queue format from our site.

A

Cool, I mean, I don't think we're gonna get the zero pull requests in q in q before before we branch, but I'm hoping we can get to uh relatively close to zero and it might make sense to either accelerate or hold polls.

A

So we can get we can. We can resolve them all before the branch. It's not required, of course, just something to think about.

A

Okay, so um we'll we'll announce that beforehand, so people have time to do it, but uh I would say tentatively, you know next sometime next week for early the week after any any concerns.

D

A

Somebody's asking: is there a roadmap for the other hypervisors mentioned in the multi-os dock dsxi? I believe that support was added in the a recent poll and zen server has been in for a long time. If I'm correct, although we don't provision zen server, you guys.

C

Want to speak to.

A

A

Vincent or adam.

D

I don't think there's anything for them. We have a support for that into the cloud, but it's really based on libya and.

C

D

The the cookbook itself only works on one suso, so it would need to be ported to ubuntu. So we could expose that in the ui upstream.

A

Cool okay and these- and these two hypervisors are actually standalone revisers. We don't we don't provision either. At this point.

D

So far for uh vmware it doesn't, but for for ipav actually does.

A

I'm sorry right, correct.

D

A

D

It needs a lot of documentation, but yeah. It does.

A

Yeah and sorry yeah, I think I'm glad you clarified that hyper-v we definitely provision as a standalone, the in the in the chat. The question is esxi and zen server and we currently don't provision either of those. Although there are requests to do it, and I know, there's community interest being.

C

A

Those things are definitely targets for the crowbar to work.

A

Cool good question, uh and I would suggest, if you're, trying to whoever asked the question, if you're trying to do this and uh bring it up through the list or chat and we'll see if we can help you, I would I'd, be very interested to see you guys if there's some zen server capability, um you know, even if it doesn't work in in the other operating systems, I'm assuming it's it's in. It's just not doc. It's just not not tested on the other platforms.

A

um We and dell does not test those those hypervisors.

A

All right so mesa brandt, we talked about mesa branch talked about havana branch, hyper v, rail. uh We are still working on on getting our ci infrastructure in place to do. Tesla polls.

A

The guys aren't here to talk specifically about it. It's still, it's still a priority for us to do it, um but since the it seems like um full, full request um manual test, support request is still paying that um the changes are less dramatic and so less risky.

A

Let's talk about the the sliding all the meeting schedules back by week, um now's a good time to talk about changes. If we want to change the these community meetings, um we've been missing a lot lately, so it's worth saying that they are they still working it's this something we want to. We want to get back into the rhythm of.

C

A

All right, my expectation, is that crib r2 stuff is heating up and having a regular cadence of design, review and planning, and coordination on chrome r2 is going to be useful and it actually looks like philosophy on crowbar overall is heating up, and we having these things, think points are really helpful. So unless somebody has a strong objection, I would prefer that we keep these cadences.

A

What I would, what I would ask is that we use today is the sprint start and that's a little bit of a change. I hope that's not disruptive to anybody.

A

So not somebody objects, we'll keep the basic cadence thursday mornings for uh community community, sprint coordination and planning coordination, and then, on the alternate weeks of design meeting where we talk about design concepts, um there's a significant backlog of uh topics that we have. So we need to start it. You know making sure we have a prioritized list and then we'll start knocking those things out.

A

I'm also planning to start doing videos and uh docs about crowbar 2 and how that operates the infrastructure, building, cookbooks and things like that. So that expect to see those things starting to come up um and I'll post as I post them I'll post through the list. So all that, all that all that pieces will come and as a reminder, the crowbar tooth system has an integrated, integrated docs, and so you will see.

A

uh That we are trying to maintain documentation in these in these stock directories. Somebody is.

A

As a crossover beep.

B

Might be, maybe somebody went on hold.

A

Yeah I like, I, don't, have visibility into who so and it's gone? No, it's not!

A

uh So I'm I'm trying to keep these docs up to date and if people have um items or or anything if they watch videos and see useful information, please, uh let's, let's make an effort as we as we start development and ramping, crowbar ii, that we we stay ahead on the dock side, certainly making sure we stay up on test and docs for the semester code. I'm doing right.

B

Is there an expectation that design discussion on the list is going to ramp up, because I haven't really seen any of that recently and I think realistically, if you want our input, that's going to be the only way to effectively get it.

A

That makes a lot of sense. Yes, uh that makes a lot of sense.

B

I mean I, I think I think it's just maybe um just a difference in in the way. The way we work, but generally you know we're very email and list oriented um and from from what I've seen in the past, some of the design calls unless you're regularly attending them and regularly sort of hands in the code.

B

It's pretty hard to follow along um some of it. um I mean less so for me, but I think for some of my colleagues who you know especially um having not been involved in the design discussions in in austin. When I, when I was there, you know I they've been there being discussion since then that I I could follow along, because I had that background understanding.

B

But anyone who hasn't been privy to those kinds of um live conversations is: is it it's just not really what's going on, whereas if you, if you, if you do it via ongoing discussions on the list where people have uh are able to sort of mentally digest the information in their own time and and to you, know cross-reference it by looking at code or um design documents or you know, schema diagrams, and that kind of thing I think that way can work.

B

But if it's just like an hour every two weeks where somebody hops on a call and tries to understand two weeks worth of of discussion, that's mostly happened on site at dell. Then I, I think, that's probably not going to be as effective.

A

Fair enough, I I one of the things that I'd like to see and we need we need some help with is to transcribe uh things that are going on in those when we have the synchronous meetings uh and create and create reasonable leap behind.

A

So I I adam I I hear what you're saying and, and we will we will step up efforts to make sure that we uh pump things out to the left and do it one of the the challenges that I wanted to see figure out how to get past. We have the same issue on our team because not everybody's, uh actually relatively few people have been involved in the crowbar ii uh design of late and so we're going to have to educate people. So they have the contacts to understand the email threats right.

A

The email threats can't can't completely rehash everything. So we have to. We have to build some background materials, so people can understand what's going on and then then have the targeted email, email conversations about it and we need help uh helping capture this that information. So people understand it.

B

Yeah I mean I appreciate it's frustrating for you guys, because you know you're sort of working together on a daily basis, face to face and you're able to make a lot of progress quickly by just you know, whiteboarding stuff and whatever, and it's it's frustrating to then have to take extra time to um you know to share that to to write it out on the list or document it and draw diagrams and so on, but like if yeah, if you, if you want the community to you, know the development community to grow outside naval.

B

I think that's that's outside dell, at least they rhyme, um then that's the only way really.

A

Yep, that's it's definitely something we need to do a better job at one of the things I feel like has happened, and I know it's hard to see this.

A

In in the whole, let's say.

A

One of the things that's been happening in the uh design work is we've actually transitioned um over into sort of the the validation of the design components. It's still almost identical to what we discussed out of, um but I realize very few people understand that so uh in the next sprint. My expectation is that our team's going to be doing an education effort and the first thing in doing that is going to be creating the docs and making sure that we have leave behind and those things are written down in a comprehensible way.

A

I won't need help to do it doing that I'll post videos of training sessions. I do and we'll create as much documentation as we can as we get and do it in a place where, where people have questions, we can have people come in and add and extend one of the things that I want to make sure we do as a community is transition, transfer those email discussions into documentation, and that's that's going to take a pretty broad spectrum effort.

B

A

I yeah adam, I know we don't do it, and so we need to build some credibility with this, but I agree I agree and support what you're requesting.

C

A

All right, I had one more note that I had this in the wrong place: anything else on on the crowbar two pieces. What I'm the biggest thing I would, I would note, is that from the status perspective, I was updating the trello.

A

We continue to make a lot of progress on the burn down.

A

It looks like the ui pieces that are are in are mostly complete.

A

Are enough to actually run the system and make it work, I'm still trying to build a complete bdd test coverage for the annealer, and that means I'm testing. Basically, I have to test everything to get there and then we've we're actually having crowbar 2, building a admin server and all the capabilities for the admin server, which includes installing shaft using the script jig and then using chef to install ntp, dns, dhcp logging and now and then the networking pieces. And then that means we can finally start building up the provisioner um and the deploy.

A

uh Sorry, the deployer first and then ultimately, the provisioner, so those pieces are actually falling into place uh right now either the polls are coming in to do that.

A

And so that's why we feel like we can start training more people, so the development effort can increase velocity.

A

That's the ver at the curb r2 status. In a nutshell,.

A

um And then the one one thing of note: it's just that we're we're trying to use docker for some of the dev environment pieces so that we can do some more um use, less resources for for some of these preliminary tests.

A

So if you're interested in that you'll start seeing that mostly in crowbar ii for the pro bar admin pieces, for we we're also talking to the docker team about havana and using docker as a hypervisor in havana and one of the things that we're considering if the integrations look good is to actually use docker as the default hypervisor for the community builds in the havana release.

A

One of the advantages of doing that is that it eliminates the need to pre-populate the the glance cache, which uh makes the glass bar clamp much smaller if you're, if you're, if you're having to transport it.

A

So if people are familiar, docker is a galaxy, it's basically an extension lxd that adds some some extra capabilities and improves the portability of containers in the lxe system. So you can treat uh docker docker containers as uh effectively portable virtual machines they're not, and one of the the things that I've been being asked about. A lot is bare metal and fair metal performance and how crowbar compares or interacts with ironic.

A

That's that that's a another topic topic for another day, but it's definitely something that I'm working out a position for, and I can certainly as a as we start to document it. People are interested. We can extend that conversation to the list because I suspect other other people who are using crowbar uh are getting similar questions.

A

Does that sound like a dialogue that we'd like to have on the list.

B

A

I'm I I'm getting a lot of enthusiasm around docker for because it we can talk about over subscription and bare metal performance uh without having to deal with all the mess of uh bare metal provisioning which kerber is built to do. But cloud deployments aren't.

C

A

All right that that was a significant agenda. um Anybody want to take time to talk about topics we didn't cover and should cover for the next week, or we can just build that up. In the background.

A

Nothing from me.

A

And it sounds like we're: we we were done so we'll get together wednesday. um One of the agent items will be that adam will do the ucs demo.

A

So and then I guess we'll figure out what else needs to be talked about by wednesday right.

A

Definitely what we we might want to do is if there's interest we can see about talking about docker on wednesday.

A

All right cool thank.

C

A

B

A