Hi Andi, Le jeudi 14 mars 2013, à 14:55 -0500, andi_a...@dell.com a écrit : > > > > Hi, > > > > If we want to be able to switch to OpenStack cookbooks from AT&T for 2.0, I > > think it makes sense to start looking at merging the PFS feature in these > > I agree that this is a big difference in functionality, but I believe that > the code changes required to integrate these capabilities is relatively > scoped, known, and pretty small.
I agree that, from a purely code perspective, the scope is not that big. But I think getting the feature accepted needs some convincing. Which is why I feel we should start pushing :-) > There are potentially bigger fish than the PFS parts, here are a few thoughts: > > * start moving towards the os-utils/openstack-common library use for > managing endpoint and uri [1], [2] So, that's a core point we need to discuss here: we can definitely do that and keep pushing convergence by working on the crowbar cookbooks. But the end result will be that we'll end up with crowbar-specific cookbooks (as in "developed for crowbar"); and we'll still be diverging from upstream ones since the common ground will be more and more limited. I very much favor the approach where we talk now with AT&T people on how to converge, and rebase things on the AT&T cookbooks. Otherwise as we progress, it will simply become more and more painful to do this. To be honest, it looks like the Grizzly branches we have in the barclamps are not moving too much right now (except for the recent quantum integration, which is great to see!). I'm not saying this as a complaint or a negative thing; on the contrary, that is totally fine and expected since we're still focusing on the core of crowbar. But this is why I feel now is a good opportunity to rebase on top of the AT&T cookbooks, before more of us move to focus on the OpenStack cookbooks. (Or maybe there's another place than the feature/grizzly/master branches to look at for progress?) > * define and agree on a common way to handle the rich physical networking > aspects currently in crowbar (e.g. bonds, vlans and friends), which I can't > see manifested in the att cookbooks (I'll ping jaypipes to start a thread on > that). e.g. in [3], there's little mention of either. This will become even > more important as we bring in Quantum. > * rationalize provider / consumer relationships. i.e. how monitoring > providers [4] get hooked in, how different coookbooks (e.g. mysql/db, > keystone) make providers available to their consumers [5] Good point, these two items are definitely things we should discuss with Jay and others. > > cookbooks, as it's probably one of the biggest difference we need to focus > > on. > > > > There's some additional work being done on the pfs, to enable using > virtualenv per component. It might be good to let that effort finish, which > should be in the next couple of weeks. Ah, good to know. What's the place where we can track the work on PFS? > > It wouldn't hurt to discuss this specifically (either on a call, or in a > > quick irc > > meeting) between the people who looked at the cookbooks from AT&T and the > > people who are familiar with the PFS feature, so we can agree on a plan. > > Agreed, preferably a voice enabled channel. But see above re: when. I'm trying to get organized a meeting with AT&T so we can discuss that kind of stuff. Would a hangout work for you? And btw, as you point out in your other answer: we should also take the opportunity to meet during the summit to discuss more on this topic. Vincent -- Les gens heureux ne sont pas pressés. _______________________________________________ Crowbar mailing list Crowbar@dell.com https://lists.us.dell.com/mailman/listinfo/crowbar For more information: http://crowbar.github.com/
