| *** openstackgerrit has quit IRC | 01:06 | |
| *** ricolin has joined #openstack-tc | 02:38 | |
| *** openstackgerrit has joined #openstack-tc | 05:52 | |
| openstackgerrit | Keiichi Hikita proposed openstack/governance master: Add qinling-dashboard under Qinling project https://review.openstack.org/591559 | 05:52 |
|---|---|---|
| openstackgerrit | Keiichi Hikita proposed openstack/governance master: Add qinling-dashboard under Qinling project https://review.openstack.org/591559 | 05:53 |
| *** ricolin has quit IRC | 06:43 | |
| *** e0ne has joined #openstack-tc | 07:18 | |
| *** jpich has joined #openstack-tc | 07:59 | |
| *** openstackstatus has quit IRC | 08:12 | |
| *** dtantsur|afk is now known as dtantsur | 08:15 | |
| *** cdent has joined #openstack-tc | 08:43 | |
| evrardjp | morning | 08:50 |
| * cdent waves | 08:52 | |
| cmurphy | o/ | 08:53 |
| *** jaosorior has quit IRC | 08:53 | |
| cdent | tc-members it's officer hours | 09:00 |
| EmilienM | hellow | 09:00 |
| cmurphy | EmilienM: this time you have to be in france | 09:01 |
| EmilienM | I'm somewhere :D | 09:01 |
| EmilienM | the city of love ;-) | 09:02 |
| evrardjp | for pidgins? | 09:03 |
| cmurphy | how are people leaning on https://review.openstack.org/590601 versus https://review.openstack.org/588644 ? | 09:07 |
| evrardjp | it seems we have a tendency to give a safety rope, so it would make sense to go towards https://review.openstack.org/590601 to me... | 09:09 |
| evrardjp | But I'd love hearing what ppl are thinking about this | 09:09 |
| evrardjp | it's giving a bad message if someone is volunteering IMO, at least it's bad if we don't give a proper explanation | 09:10 |
| evrardjp | again, just my personal opinion | 09:10 |
| cdent | cmurphy: I can't decide. I agree that having a volunteer is a good sign and meaningful. But on the other hand at some point it feels like (and this phrasing is a bit stronger than I mean and not meant to reflect on any projects) we need to cut the dead weight | 09:14 |
| evrardjp | cdent: I agree on the dead weight, but here there is someone that's willing to step in | 09:17 |
| cdent | yeah, thus the indecision | 09:17 |
| cdent | the problem is that searchlight has been effectively dead for about a year | 09:17 |
| evrardjp | that breaks my heart to say "no we have decided to kill this, continue this on your own if you want" | 09:17 |
| evrardjp | cdent: that's true | 09:18 |
| cdent | and "not governed" is not the same as "killed" | 09:18 |
| evrardjp | but if it's more than a year it doesn't change | 09:18 |
| evrardjp | that's why my question yesterday: what's the risk of having "not governed" projects | 09:19 |
| evrardjp | I think this was well known when you see: https://github.com/openstack/governance/blob/master/reference/projects.yaml#L3256 | 09:20 |
| evrardjp | I think the governance repo is lacking a "temporal" aspect that might be used. | 09:21 |
| evrardjp | anyway, sorry to have intervened, go backs to lurking mode. | 09:22 |
| cdent | this stuff is open to everyone. that's kind of the point. no need to lurk. | 09:24 |
| *** openstackstatus has joined #openstack-tc | 09:42 | |
| *** ChanServ sets mode: +v openstackstatus | 09:42 | |
| *** jaosorior has joined #openstack-tc | 10:51 | |
| *** jaosorior has quit IRC | 11:17 | |
| *** jaosorior has joined #openstack-tc | 11:17 | |
| *** rosmaita has joined #openstack-tc | 11:33 | |
| smcginnis | I actually forgot we tagged searchlight as maintenance-mode. | 12:24 |
| smcginnis | I wonder if we need to change the way we do some things for projects with that tag. | 12:25 |
| smcginnis | No expected releases every cycle, no PTL election, PTL stays the contact point for the project until they hand off to someone else or we find something needs attention and they are not able or willing to provide it. | 12:26 |
| cdent | s/things.*/things/ | 12:26 |
| smcginnis | Seems like a maintenance-mode project really should have different expectations and requirements than a normal active one. | 12:27 |
| smcginnis | cdent: Hah, fair enough. | 12:27 |
| *** tellesnobrega has joined #openstack-tc | 12:27 | |
| cdent | I think if we don't require the ptl election and releases, both, then we make it easy for the project to slip under the radar. The ptl election is sort of like a "yeah, I'm still here and paying attention check" | 12:28 |
| cdent | I agree that requiring releases for something that is "stable" is odd | 12:28 |
| cmurphy | ++ I don't think being able to get in contact with the PTL and reminding them to submit a change to the elections repo once every six months should be too much to ask | 12:29 |
| smcginnis | True | 12:29 |
| cdent | It might, however, make sense for an individual to be the stable-maintenance ptl for multiple projects | 12:32 |
| smcginnis | That could work. | 12:35 |
| *** jaosorior has quit IRC | 12:39 | |
| *** zaneb has joined #openstack-tc | 13:22 | |
| TheJulia | cdent: That is kind of what I was thinking in the back of my head would be ideal | 13:33 |
| *** jaosorior has joined #openstack-tc | 13:44 | |
| dhellmann | does anything about our structure prevent that today? someone could just sign up, right? | 14:02 |
| jroll | the only catch is they'd need a commit to the project | 14:03 |
| cdent | jroll: I think that's the crux of the biscuit, yeah. which is why naming it something slightly different might help | 14:04 |
| jroll | yep | 14:05 |
| cdent | dhellmann: and underlying the idea is basically a sense of an even further back step: members of the TC (as licensed upstream over-committers) become designated stable-maint ptls | 14:05 |
| cdent | which I do _not_ think we should do | 14:05 |
| cdent | but highlights what it could all mean | 14:05 |
| jroll | the different name might help with external perception too - this person won't be an expert on searchlight/etc, but can help with maintenance and point people | 14:05 |
| jroll | or s/perception/understanding/ | 14:05 |
| * cdent nods at jroll | 14:06 | |
| cdent | pabelanger: have you made any contact wiith masakari folk? | 14:06 |
| cdent | i don't want to repeat effort if you have | 14:06 |
| pabelanger | cdent: I have not | 14:11 |
| pabelanger | but can if you'd like | 14:11 |
| cdent | pabelanger: if you're willing and able that would be awesome | 14:12 |
| pabelanger | sure, will send out an email today | 14:13 |
| cdent | cool, thanks | 14:14 |
| *** ricolin has joined #openstack-tc | 14:15 | |
| dhellmann | this still feels like one of those cases where we want to attach a special name to something that people could already do | 14:18 |
| dhellmann | I'm not sure that's bad, it just seems surprising that it came up | 14:19 |
| cdent | people may be scared of being held responsible for something in a way that implies more than they can do | 14:27 |
| cdent | There's a different expectation between a teacher and a substitute teacher | 14:28 |
| smcginnis | Good analogy. | 14:31 |
| * cdent is full of 'em | 14:32 | |
| dhellmann | maybe rather than having 1 person maintain multiple projects as a sub, we want to rehome those projects to a team with that as its mission | 14:32 |
| dhellmann | that would be somewhat like the oslo team does | 14:32 |
| dhellmann | tc-members: please take a few minutes to go through the open reviews, express your opinion on https://review.openstack.org/#/c/590790/, and confirm the PTL appointments | 14:37 |
| EmilienM | dhellmann: ack | 14:37 |
| EmilienM | (ah, done already) | 14:38 |
| cdent | zaneb: I assume you saw jay's mulligan? I was hoping for openstack v2, not nova v2, but despite that it still has some interesting things to think about. | 14:44 |
| zaneb | cdent: I only read part 1 so far | 14:44 |
| *** annabelleB has joined #openstack-tc | 14:44 | |
| cdent | how very disciplined(?) of you :() | 14:45 |
| zaneb | I saw it at like 11pm on Friday and I was on PTO yesterday | 14:45 |
| zaneb | was slightly confused about why, if you wanted to start a project with no vendors, no conferences, and no scope outside of managing virtual machines, you would start with a project that has thousands of people doing all of those things and then try to kick them out | 14:46 |
| zaneb | but I am excited for part 2 | 14:46 |
| dhellmann | the company I worked for before I worked on OpenStack built Mulligan (though we didn't call it that). There were about 8 of us, and we built something much like what he describes. We did no marketing, so between that and the fact that it met no one's real needs we had no customers. They're out of business now. | 14:48 |
| dhellmann | So, I'm not sure any of that is really what I would call a winning strategy. | 14:48 |
| cdent | perhaps useful as a Gedankenexperiment | 14:49 |
| zaneb | dhellmann: almost as if all those vendors and conferences contributed something of value after all | 14:49 |
| dhellmann | zaneb : I do like living in a house. | 14:49 |
| dhellmann | cdent : if it was "we should fix some architectural issues" then yes. But that's not really how I read it. | 14:49 |
| cdent | dhellmann: I dunno. If someone lays out a model for a different thing, that still provides by potential for stuff worth stealing. It's fertile as ideas, but not ncessarily as actions. | 14:50 |
| dhellmann | which ideas would you take, then? | 14:51 |
| cdent | Despite what jay says in his posts about do instead of write I think we _way_ under-write and reflect and digest | 14:51 |
| cdent | the biggest thing I think worth thinking about is not at all original is jay's thing: use etcd and watchers thereof as a way of managing intent | 14:52 |
| cdent | s/is/in/ | 14:52 |
| zaneb | UGH SPOILERS | 14:52 |
| cdent | And, in any case, it's the ideas that come as secondary thoughts from reading others' ideas that matter | 14:52 |
| cdent | stimulation, even to do the opposite, is good | 14:52 |
| cdent | zaneb: the butler did it | 14:53 |
| cdent | Mind you, I'm no trying to indicate support for jay's proposals. I don't think it's practical or practicable | 14:54 |
| cdent | but it is food | 14:54 |
| dhellmann | it feels to me like jay misses randy bias' "contributions" to the community | 14:55 |
| dhellmann | I haven't read the 2nd post yet though, so maybe there's more substance there | 14:55 |
| scas | it's a great thought exercise, and is in-line with the open source mindset. i think if applied, it would send all sorts of signals, not all intended | 14:58 |
| fungi | still reading scrollback, but i like the idea of not requiring releases for maintenance-mode projects (maybe that tag should come with its own release model of the same name?) and allowing any atc to volunteer for ptl even if they're not an apc (the extreme low number of commits could mean there was no particular reason for a developer who cares about the project to need to get a commit merged to it) | 14:58 |
| fungi | or put another way, if a project has almost no commits then the pool of candidates for ptl is remarkably constranied | 14:59 |
| dhellmann | we already have a release model for projects that don't do a lot of releases, or don't do a release every cycle. | 14:59 |
| scas | <-- | 14:59 |
| scas | chef is in one of those grey areas of release models. one 'coordinated' release a cycle is what i strive for, with incremental updates to the stable branches as folks consume the next release | 15:01 |
| scas | in practice, without a team all working toward that goal, work really doesn't start until packages are GA | 15:02 |
| scas | of course, the word 'release' is somewhat loaded. what i consider 'released' is 'on supermarket', which is different than someone else's interpretation | 15:04 |
| *** e0ne has quit IRC | 15:16 | |
| persia | dhellmann: Maybe when projects go into maintenance mode, part of that process should be a review of the release model with the release team? | 15:19 |
| dhellmann | sure, that makes sense | 15:20 |
| *** e0ne has joined #openstack-tc | 15:20 | |
| openstackgerrit | Doug Hellmann proposed openstack/constellations master: import zuul job settings from project-config https://review.openstack.org/591739 | 15:23 |
| openstackgerrit | Doug Hellmann proposed openstack/constellations master: switch documentation job to new PTI https://review.openstack.org/591740 | 15:23 |
| *** jungleboyj has joined #openstack-tc | 15:32 | |
| jaypipes | dhellmann: at the end of the second blog post, I'm pretty clear that what I'm describing isn't "OpenStack v2". It's a completely different thing, with little to no connection to openstack other than learning some lessons in what not to do by building a giant marketing hype machine. | 15:36 |
| dhellmann | I haven't had a chance to read that yet. | 15:37 |
| jaypipes | dhellmann: it was originally about fixing architectural issues. but, sorry, I just don't think openstack is able to overcome its existing weight and change fundamental directions from an architectural perspective. | 15:39 |
| jaypipes | dhellmann: too much legacy. too much vendor input. too little focus on a small footprint and small scope. | 15:40 |
| jaypipes | perhaps Randy Bias hit on too many truthful points... | 15:43 |
| *** e0ne has quit IRC | 15:52 | |
| *** cdent has quit IRC | 15:58 | |
| *** jpich has quit IRC | 16:38 | |
| *** ricolin has quit IRC | 17:05 | |
| *** rosmaita has quit IRC | 17:05 | |
| openstackgerrit | Doug Hellmann proposed openstack/constellations master: use python3 in tox https://review.openstack.org/591786 | 17:09 |
| *** zaneb has quit IRC | 17:45 | |
| *** eandersson has joined #openstack-tc | 17:46 | |
| *** dtantsur is now known as dtantsur|afk | 18:15 | |
| *** annabelleB has quit IRC | 18:25 | |
| *** annabelleB has joined #openstack-tc | 18:29 | |
| *** diablo_rojo has joined #openstack-tc | 18:46 | |
| *** annabelleB has quit IRC | 19:06 | |
| *** zaneb has joined #openstack-tc | 19:22 | |
| *** e0ne has joined #openstack-tc | 19:23 | |
| *** e0ne has quit IRC | 19:26 | |
| *** e0ne has joined #openstack-tc | 19:27 | |
| *** annabelleB has joined #openstack-tc | 19:37 | |
| zaneb | ok I finished part 2 | 19:44 |
| zaneb | jaypipes: I'm impressed that you actually managed to require more external dependencies than Nova v1 has :P | 19:44 |
| *** jaosorior has quit IRC | 19:45 | |
| *** cdent has joined #openstack-tc | 19:53 | |
| jaypipes | zaneb: yes, three is indeed onerous. | 19:54 |
| zaneb | jaypipes: if I were to write a similar blog post I would use FoundationDB for everything | 19:55 |
| cdent | if _I_ were to write a similar blog post I would persist as little as possible | 19:58 |
| *** jaosorior has joined #openstack-tc | 19:58 | |
| jaypipes | zaneb: not having SQL for relational data just means you end up writing joins in memory. | 19:59 |
| jaypipes | zaneb: but it's fine for certain types of data. | 19:59 |
| jaypipes | zaneb: also, not being able to do aggregate querying on highly relational data like usage tables is just choosing the wrong tool for the job, IMO. | 20:00 |
| zaneb | jaypipes: they reportedly had a layer (unfortunately not open sourced) that was on-the-wire-compatible with postgres. in theory there's nothing you can do in sql that you can't do in FoundatoinDB | 20:00 |
| jaypipes | zaneb: why not cockroachdb if you want distributed SQL? | 20:00 |
| jaypipes | it's PG-compatible (mostly) | 20:01 |
| zaneb | because you can also do simpler stuff without needing to implement a full-on SQL layer | 20:01 |
| zaneb | like queues | 20:02 |
| zaneb | which don't go well in an SQL DB in my experience ;) | 20:02 |
| jaypipes | zaneb: pretty sure I've never recommended writing a queue using a SQL db. | 20:03 |
| jaypipes | cdent: you gotta put data somewhere, right? :) | 20:04 |
| zaneb | jaypipes: right, but I'm saying FoundationDB could potentially replace all of the DB, queue, and etcd components. whereas cockroach can probably only replace the DB part which leaves us back where we started | 20:04 |
| jaypipes | zaneb: it's not a "DB" I need. It's SQL. | 20:04 |
| jaypipes | zaneb: if all I need is a "DB", I'll use etcd for it. | 20:05 |
| zaneb | ok, the SQL, key-value store, and queue components | 20:05 |
| jaypipes | zaneb: i.e. I want to use the right tool for the job. | 20:05 |
| jaypipes | zaneb: I would imagine that if you used foundationdb as your only data store, you'd run into the same kinds of square-peg-round-hole problems that *both* Nova and k8s have exhibited. | 20:06 |
| jaypipes | zaneb: Nova in the case of being too reliant on a relational SQL DB for everything and k8s for thinking a k/v store (transactional even) is the right tool when you need to perform aggregate querying and calculations. | 20:07 |
| zaneb | maybe | 20:08 |
| zaneb | it'd certainly be a lot easier to argue for if they'd open sourced the SQL layer | 20:08 |
| jaypipes | zaneb: but if you're telling me foundationDB "has it all", then I'd consider it. but from what I can gather, it's not a real SQL database. | 20:08 |
| zaneb | but everything I've read suggests to me that this is basically what AWS is doing with Dynamo | 20:09 |
| jaypipes | zaneb: sorry, not following you... by "that this" you mean putting a SQL layer on top of a distributed K/V store? | 20:11 |
| zaneb | don't know about an SQL layer, but it appears to me that all of AWS runs on Dynamo | 20:12 |
| jaypipes | zaneb: because that ain't new by a long shot :) facebook's been doing that for years with rocksdb and myrocks (http://myrocks.io/) | 20:13 |
| jaypipes | zaneb: oh, you mean AWS has been dogfooding itself with Dynamo.. yes, true dat | 20:14 |
| zaneb | it's a confusing topic to discuss because AWS also now has a product called DynamoDB that makes (parts of?) Dynamo available to end users | 20:16 |
| cdent | jaypipes: data either goes in ram/cache or $some_layer_above that cares | 20:32 |
| cdent | jaypipes: also, there would be infinite resource classes | 20:32 |
| jaypipes | cdent: heh | 20:36 |
| *** jaosorior has quit IRC | 20:56 | |
| *** diablo_rojo has quit IRC | 21:05 | |
| *** annabelleB has quit IRC | 21:23 | |
| *** annabelleB has joined #openstack-tc | 21:38 | |
| *** cdent has quit IRC | 21:59 | |
| *** e0ne has quit IRC | 22:39 | |
| *** annabelleB has quit IRC | 23:54 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!