Thursday, 2016-01-28

*** dims_ has quit IRC00:55
*** amotoki has quit IRC02:59
*** amotoki has joined #openstack-release03:27
*** amotoki has quit IRC03:32
*** yarkot has joined #openstack-release03:45
*** amotoki has joined #openstack-release03:49
*** amotoki has quit IRC04:03
*** bdemers has quit IRC04:07
*** bdemers_ has joined #openstack-release04:07
*** amotoki has joined #openstack-release04:20
*** amotoki has quit IRC08:07
*** amotoki has joined #openstack-release08:16
*** amotoki has quit IRC08:18
*** amotoki has joined #openstack-release08:42
*** mriedem has joined #openstack-release08:57
*** dims has joined #openstack-release11:07
*** doug-fish has joined #openstack-release11:49
*** amotoki_ has joined #openstack-release12:41
*** amotoki has quit IRC12:43
*** dims has quit IRC12:47
*** amotoki_ has quit IRC14:16
*** sigmavirus24_awa is now known as sigmavirus2415:08
*** dims has joined #openstack-release15:08
*** rhochmuth has joined #openstack-release15:58
dhellmannhi, rhochmuth16:05
dhellmannso let's talk about monasca's release model, and then you had some build issues you wanted to talk about, too16:06
* dhellmann looks for his notes16:06
rhochmuthsure, although i think we are over our build issues16:06
dhellmannok, cool16:06
rhochmuthwe figured it out16:06
rhochmuthso, probably jsut release model16:06
dhellmannlooking at
dhellmannttx had a note that you're listed as "cycle-with-milestones" but you've been doing intermediary (non-beta) releases16:07
dhellmannalso that apparently different components of the same "deliverable" are tagged with different versions16:08
dhellmannthe latter one just means we need different deliverables defined16:08
dhellmannand the former means we should change your model to cycle-with-intermediary instead16:08
rhochmuthwe've been tagging lot's of "releases"16:08
dhellmannbut, before we do that, I wanted to make sure you understood the implications16:08
dhellmannand the difference between the two16:08
dhellmannthe milestone model implies you'll have one official release per cycle, with betas and release candidates leading up to that16:09
dhellmannthe intermediary model sounds more like what you're doing, but also implies that you'll have one *final* release used to create a stable branch for the cycle16:09
rhochmuththe latter sounds like something that we could do16:10
dhellmannthere's also an independent model that lets you release at any time, but projects using that tag are not part of the release cycle and can't claim to be part of "mitaka" so I don't think that's what you want16:10
dhellmannwe usually reserve that for things outside of the real release, like tools16:10
dhellmannok, so we can change the release model designation easily enough16:10
rhochmuthi think if we can do intermediary releases, but also have a formal releast that woudl be best16:10
dhellmannso, deliverables16:11
dhellmanna deliverable may come from multiple repositories, but the expectation is that they are tagged together with the same version number at each release16:11
rhochmuthok, that sounds reasonable16:11
rhochmuthwe did something like that back in may with at 2015.1 tag, or something similar16:11
dhellmannso does the existing "monasca" deliverable need to be split up?16:12
*** amotoki has joined #openstack-release16:12
rhochmuththere are multiple repos, but we could apply the same tag to all of them16:12
rhochmuthin general, master branch is very high quality, so we can basially tag at anytime16:13
dhellmannhere's what I see now:
dhellmannso you have things with completely different versions, and at least one that isn't released at all16:14
dhellmannit's fine to split them up into separate deliverables if you want to version them separately -- and at this point that's probably the most sane thing to do otherwise the versions will jump around a lot to get them into sync16:15
rhochmuthso, we've been using a tag that looks like 1.1.16, or something similar, for all of the intermediate releases16:15
rhochmuthand then we had applied a tag of 2015.1 back in May16:16
dhellmannit's not the style of the versions I mean16:16
rhochmuththat was supposed to be more of a "formal" release16:16
dhellmannok, that's definitely not how we do versioning16:16
dhellmannyou should no longer use date-based versions at all16:16
dhellmannwe've moved everything to semver:
rhochmuthso, is there a specific tag that we should be using for releases, like mitaka16:18
dhellmannswitching back and forth causes issues for packagers, since the version  numbers jump up and down16:18
rhochmuthso, we have several repos.16:18
dhellmannwe started the other projects on semver by counting how many releases they had already done, adding 1, and using that as the major version16:18
rhochmuthmost of them use a 1.0.X format16:18
rhochmuthbut they are our of sync16:19
dhellmannyep, those are good16:19
rhochmuthwe tag each independtly16:19
dhellmannok, they don't need to be in sync16:19
dhellmannbut if they are out of sync, they are not part of the same deliverable16:19
dhellmannthat's ok -- it's just documentation so people understand what repos are closely tied together16:19
dhellmannsince I don't see the same tag on any 2 repos, I think you've got one repo per deliverable16:20
dhellmanneach deliverable can have its own version16:20
dhellmannat the end of mitaka, we'll document which final version is the stable version for mitaka16:20
rhochmuthso, if we apply a new tag for each repo at the same time, then that could be a release16:20
dhellmannthen in newton you'll raise the minor version number (going from 1.0 to 1.1) at least to leave room for stable releases16:20
dhellmannreleases, deliverables, and repos are different but related things16:21
dhellmannyou release a deliverable by tagging all of the repos that make up that deliverable16:21
dhellmannthe deliverable has a version, and the version is applied to each repo at the same time16:21
dhellmannif you release parts of the system independently, they don't need to be part of the same deliverable16:21
dhellmannhere are the deliverable definitions for other projects in mitaka:
dhellmannthe astara project has several independent components, so those are listed as separate deliverables16:23
dhellmannneutron has several repos in one deliverable, so that's one file16:23
dhellmannby the way, if you want to document which versions of monasca things are part of mitaka, you should do that in the openstack/releases repo16:24
rhochmuthi think we could follow the neutron model16:25
dhellmannin that case, you'd need to sync the version numbers of the things that go into the deliverable16:25
rhochmuthi see, they have all the same major version16:26
dhellmannthey all have exactly the same version16:26
dhellmannin fact, the way they *get* that version is by me running a script that reads the file in the releases repo16:26
rhochmuthhmmm, i didn't quite understand16:26
rhochmuthso astra16:26
dhellmannastara looks like they're doing similar version numbers right now, even though they treat the parts as separate16:27
dhellmannthat implies they expect to possibly do stable releases of the components independently16:27
dhellmannwhich makes sense for them, because the appliance image may change but the service code doesn't need to16:27
rhochmuthall their versions are like,
dhellmannso they are future-proofing their deliverable definitions to allow for separate releases later16:28
dhellmannglance and glance-store may be a better example16:28
dhellmannor the ironic deliverables16:28
dhellmannsome of the ironic components don't even have the same major version number16:29
dhellmannsome have been released a bunch of times, some only once16:29
rhochmuthso, using ironic as a model, that seems to fit what we've been doing so far16:29
dhellmannright, and that's fine -- we just need to split them up in the governance repo to reflect that16:30
dhellmannand eventually we'll need for you to record the releases in the releases repo using the same structure16:30
dhellmannthat is, the same set of deliverables and repos organized in the same way16:30
dhellmannit would be good if you could start doing that soon, so we're up to date by M3 and prepared for the end of cycle releases16:31
dhellmannrhochmuth : make sense?16:31
rhochmuthiyes, absolutely16:31
rhochmuthi think i need to dig in deeper on this16:32
dhellmannrhochmuth : ok, I can work on the patches and let you +1, or you can do them and I can +1.16:32
rhochmuthnot sure i've quite conceptualized yet16:32
dhellmannyeah, it's just an organizational layering thing16:32
rhochmuthlet me take a look at the project and go thorugh these notes16:32
dhellmannwe should also talk about the independent projects16:33
dhellmannrhochmuth: monasca-statsd and monasca-ceilometer are release:independent, which means they will not be considered part of mitaka. Is that what you want for this cycle?16:34
rhochmuthi think that is acceptable16:34
dhellmannok, good, just making sure16:35
rhochmuthalthough, i woudl like to discuss it with some folks16:35
rhochmuththat are more involved with monasca-ceilometer16:35
rhochmuthit may make sense to sync-up16:35
dhellmannsure. I would like to have this nailed down by the end of next week. The freeze date for changing models was actually *last* week, but we didn't catch  up with you in time.16:35
rhochmuthas monasca-ceilometer is dependent on ceilometer and monasca16:35
dhellmannif it lags behind, an independent model may make sense16:35
rhochmuthok, i'll start digging in16:36
dhellmannthey can't really be part of the release if they have to wait for the release...16:36
dhellmannok, good. please ping me when you're ready, either with links to the reviews in the governance and releases repo, or with notes so I can help you prepare those patches16:36
rhochmuthmonasca-statsd could be an independent16:36
rhochmuthok, thanks16:36
dhellmannand of course, any time in between if you have questions, my bouncer is always subscribed to this room16:37
*** sigmavirus24 is now known as sigmavirus24_awa17:16
*** mriedem has quit IRC17:16
openstackgerritGraham Hayes proposed openstack/releases: Release python-designateclient 1.6.0
*** bdemers has joined #openstack-release18:28
*** bdemers_ has quit IRC18:31
*** sigmavirus24_awa is now known as sigmavirus2418:42
openstackgerritJim Rollenhagen proposed openstack/releases: python-ironicclient 1.1.0
openstackgerritJoshua Harlow proposed openstack/releases: Futurist 0.10.0 mitaka release
openstackgerritJoshua Harlow proposed openstack/releases: Futurist 0.10.0 mitaka release
openstackgerritMerged openstack/releases: Futurist 0.10.0 mitaka release
*** bnemec has quit IRC20:18
*** bnemec has joined #openstack-release20:25
*** Guest94234 is now known as jgriffith23:14
*** gordc has quit IRC23:38
*** sigmavirus24 is now known as sigmavirus24_awa23:56
*** amotoki has quit IRC23:56

Generated by 2.14.0 by Marius Gedminas - find it at!