| *** robcresswell has quit IRC | 00:12 | |
| prometheanfire | yarp | 00:16 |
|---|---|---|
| *** lbragstad has quit IRC | 00:57 | |
| *** cjloader has quit IRC | 01:26 | |
| *** robcresswell has joined #openstack-requirements | 02:09 | |
| *** lbragstad has joined #openstack-requirements | 02:14 | |
| openstackgerrit | Merged openstack/requirements master: update constraint for glance_store to new release 0.23.0 https://review.openstack.org/535564 | 03:00 |
| *** coolsvap has joined #openstack-requirements | 04:15 | |
| openstackgerrit | lei zhang proposed openstack/requirements master: Raise minimum version of openstackdocstheme to 1.18.1 https://review.openstack.org/536670 | 04:16 |
| *** cjloader has joined #openstack-requirements | 04:34 | |
| *** cjloader has quit IRC | 04:36 | |
| *** cjloader has joined #openstack-requirements | 04:36 | |
| *** cjloader has quit IRC | 04:53 | |
| *** cjloader_ has joined #openstack-requirements | 04:55 | |
| *** cjloader_ has quit IRC | 04:56 | |
| *** cjloader has joined #openstack-requirements | 05:11 | |
| *** cjloader has quit IRC | 05:20 | |
| *** cjloader has joined #openstack-requirements | 05:21 | |
| *** amotoki has quit IRC | 06:03 | |
| *** amotoki has joined #openstack-requirements | 06:04 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: Updated from generate-constraints https://review.openstack.org/536217 | 06:21 |
| *** cjloader has quit IRC | 06:34 | |
| openstackgerrit | Dirk Mueller proposed openstack/requirements master: upgrade upper constraints: exabgp 4.0.4 https://review.openstack.org/536141 | 06:43 |
| openstackgerrit | Dirk Mueller proposed openstack/requirements master: update constraint for os-brick to new release 2.3.0 https://review.openstack.org/535526 | 06:47 |
| openstackgerrit | Thomas Morin proposed openstack/requirements master: upgrade global requirements: exabgp 4.0.4 https://review.openstack.org/536142 | 06:49 |
| openstackgerrit | Merged openstack/requirements master: update constraint for python-muranoclient to new release 1.0.1 https://review.openstack.org/536510 | 07:03 |
| *** jrist has quit IRC | 07:15 | |
| *** ralonsoh has joined #openstack-requirements | 07:51 | |
| *** jrist has joined #openstack-requirements | 07:53 | |
| *** jpich has joined #openstack-requirements | 09:03 | |
| openstackgerrit | Thomas Morin proposed openstack/requirements master: upgrade global requirements: exabgp 4.0.4 https://review.openstack.org/536142 | 09:20 |
| openstackgerrit | Dirk Mueller proposed openstack/requirements master: Updated from generate-constraints https://review.openstack.org/535173 | 09:20 |
| openstackgerrit | Dirk Mueller proposed openstack/requirements master: Updated from generate-constraints https://review.openstack.org/536217 | 09:21 |
| openstackgerrit | Dirk Mueller proposed openstack/requirements master: Remove msgpack-python https://review.openstack.org/533606 | 09:27 |
| *** ralonsoh_ has joined #openstack-requirements | 10:31 | |
| *** ralonsoh has quit IRC | 10:34 | |
| *** ralonsoh__ has joined #openstack-requirements | 10:58 | |
| *** ralonsoh_ has quit IRC | 11:01 | |
| openstackgerrit | Claudiu Belu proposed openstack/requirements master: Bumps os-win to 3.0.0 https://review.openstack.org/536774 | 11:19 |
| openstackgerrit | Merged openstack/requirements master: add argument error handling to grep-all.sh https://review.openstack.org/536605 | 12:15 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for python-solumclient to new release 2.6.1 https://review.openstack.org/536793 | 12:32 |
| *** ralonsoh__ is now known as ralonsoh | 12:55 | |
| *** dims has quit IRC | 13:22 | |
| *** dims has joined #openstack-requirements | 13:25 | |
| *** cjloader has joined #openstack-requirements | 13:43 | |
| *** edmondsw has joined #openstack-requirements | 13:49 | |
| *** cjloader has quit IRC | 14:00 | |
| dhellmann | prometheanfire : I think a while back you gave me a link to where you're tracking work on the "stop syncing requirements settings" project but I can't find it. Was there an etherpad? | 14:27 |
| prometheanfire | dhellmann: just the bug at this point | 14:36 |
| dhellmann | https://bugs.launchpad.net/openstack-requirements/+bug/1719009 right? | 14:36 |
| openstack | Launchpad bug 1719009 in OpenStack Global Requirements "per project requirements" [High,New] | 14:36 |
| dhellmann | I was looking for a bookmark to an etherpad, not a bug | 14:37 |
| dhellmann | yay, launchpad is giving timeout errors updating the bug | 14:38 |
| dhellmann | prometheanfire : why do we need a separate test to ensure the lower bounds are available? | 14:39 |
| prometheanfire | dhellmann: we need somthing to test against, just like we have upper-constraints.txt | 14:46 |
| dhellmann | prometheanfire : test what, though? lower-bounds will be different for all projects. we can't constrain them to the same list. any lower bound constraint needs to be in the project repo. | 14:46 |
| dhellmann | https://etherpad.openstack.org/p/ocata-requirements-notes has the original notes on this | 14:46 |
| prometheanfire | the lower-reqs in requirements.txt is not what would be tested | 14:47 |
| edmondsw | prometheanfire are you trying to get the networkx requirements cap issue worked out for queens or rocky? | 14:48 |
| dhellmann | comment #2 on that bug talks about taking "gr lower bounds an apply them as a constraint" | 14:48 |
| prometheanfire | https://etherpad.openstack.org/p/queens-PTG-requirements-divergent | 14:48 |
| prometheanfire | edmondsw: I don't think we have time for it this cycle | 14:48 |
| prometheanfire | afk, driving now | 14:48 |
| *** ralonsoh_ has joined #openstack-requirements | 14:48 | |
| edmondsw | prometheanfire ok, that's what I was thinking but wanted to confirm | 14:48 |
| edmondsw | prometheanfire so there's no rush for a new version of pypowervm then | 14:49 |
| dhellmann | prometheanfire : back during ocata we talked about removing version information from the global-requirements.txt file entirely because it would not be needed any more if we were not syncing lower bounds. We would retain the constraints file, and we said we might optionally retain exclusions to make generating the constraint simpler. So I'm confused by any mention of lower bounds coming from the global file. | 14:50 |
| *** ralonsoh has quit IRC | 14:51 | |
| *** cjloader has joined #openstack-requirements | 14:54 | |
| prometheanfire | dhellmann: it doesn't, that's a parallel effort by dirk | 15:05 |
| dhellmann | prometheanfire : ok. so what's the next step in the process of actually dropping the sync job? I would like to contribute to moving that along, if I can. | 15:05 |
| prometheanfire | dhellmann: next step, is to make a tool to generate a lower-constraints.txt from a projects requirements.txt and either use that directly in a lower-constraints test | 15:07 |
| prometheanfire | or commit it to the project | 15:07 |
| dhellmann | prometheanfire : not all projects are going to want to run those tests. is that a blocker for turning off the job that syncs dependency settings? | 15:10 |
| dhellmann | we'd probably want to commit the file to the tree, so it can be controlled | 15:10 |
| dhellmann | we probably don't want to turn that off until all of the queens branches are created anyway, so I'm just trying to understand the sequencing of the tasks | 15:12 |
| *** ralonsoh__ has joined #openstack-requirements | 15:17 | |
| prometheanfire | dhellmann: it'd be opt in, per .zuul stuff | 15:20 |
| dhellmann | sure. like I said, I'm trying to understand if that's actually blocking taking any other steps. | 15:20 |
| *** ralonsoh_ has quit IRC | 15:20 | |
| prometheanfire | ya, I'm just doing 2-3 things right now | 15:22 |
| prometheanfire | including trying to see why zoom is core dumping again | 15:23 |
| prometheanfire | but basically, it's like this (from memory) | 15:24 |
| prometheanfire | 1. make a tool in the reqs repo that converts a requirements.txt file to a lower-constraints file (this is why we have lower bounds everywhere now) | 15:24 |
| prometheanfire | 2. make a test in a project that will take said tool and run unit tests using the lower-constraints.txt that it spits out | 15:25 |
| prometheanfire | 3. make that test voting | 15:25 |
| prometheanfire | as a separate stream | 15:26 |
| prometheanfire | 1. make it an option for the requirements update bot to only push excluded versions | 15:26 |
| prometheanfire | back to the original stream | 15:27 |
| dhellmann | ugh, zoom | 15:27 |
| prometheanfire | 4 once voting, have project consume new requirements bot update method of excluded versions only | 15:27 |
| prometheanfire | uh huh | 15:27 |
| prometheanfire | yo, I hear you like vendoring | 15:27 |
| prometheanfire | also, somone needs to tell them how to package | 15:28 |
| prometheanfire | Qt WebEngine ICU data not found at /home/levi/Qt5.9.2/5.9.2/gcc_64/resources. Trying parent directory... | 15:28 |
| prometheanfire | someone tell levi | 15:28 |
| dhellmann | so, what's the harm in just stopping the sync right off, while we work on those other bits? we have no less testing under that scheme than we do now, right? | 15:28 |
| dhellmann | wow | 15:28 |
| prometheanfire | they still need to test upper-constraints and consume exclusions for co-installability | 15:29 |
| prometheanfire | particularly exclusions for co-install | 15:29 |
| dhellmann | ah, true | 15:29 |
| dhellmann | so that's really the next thing we need: the separate script or mode or whatever that syncs only the exclusions | 15:29 |
| dhellmann | actually hang on | 15:30 |
| dhellmann | co-installation is guaranteed by the global constraints list. | 15:30 |
| dhellmann | exclusions are a convenience to keep that list straight. | 15:30 |
| dhellmann | just because a version breaks project A doesn't mean project B can't use it, as long as it doesn't break B as well | 15:30 |
| dhellmann | so I'm not sure we do need to sync the exclusions. | 15:31 |
| prometheanfire | you are right, we don't | 15:31 |
| prometheanfire | so just the 4 steps, without the separate step 1 | 15:31 |
| dhellmann | I forget how the job that tests requirements changes in each project tree works, but we need to update that to only look for the names in the g-r list | 15:32 |
| prometheanfire | the problem we have is secondary, indirect deps in a project | 15:33 |
| prometheanfire | project depends on libfoo, libfoo depends on libbar | 15:33 |
| prometheanfire | project only lists libfoo | 15:33 |
| prometheanfire | so libbar's lower-constraint is what? 0.0.1_alpha1 | 15:33 |
| dirk | prometheanfire: we can capture the version deps in a separate file and generate things.. it would simplify things | 15:34 |
| prometheanfire | that's why we need to store a lower-requirements.txt iirc (they become the project's lower bounds for stuff like that | 15:34 |
| prometheanfire | dirk: pronouns confuse me :P | 15:34 |
| dhellmann | prometheanfire : ah, so lower-requirements.txt includes indirect dependencies? | 15:34 |
| prometheanfire | lower-constraints.txt includes indirect dependencies | 15:35 |
| dirk | prometheanfire: sorry: I also spent no real effort into reading the backlog ;-) | 15:35 |
| prometheanfire | lower-requirements.txt doesn't exist | 15:36 |
| dhellmann | so I think to start we just let lower-constraints.txt hold the highest version of indirect dependencies that work. If folks want to adjust those settings, they can do it by hand. | 15:37 |
| prometheanfire | lower-requirements.txt can be dynamically generated from lower-constraints.txt and requiremnts.txt, with packages in lower-constraints and NOT in requirements.txt being the lower bounds of those package versions in lower-requirements.txt | 15:37 |
| prometheanfire | basically, yes, that | 15:38 |
| dhellmann | you keep saying lower-requirements.txt and lower-constraints.txt -- are those different things? | 15:38 |
| prometheanfire | yes | 15:38 |
| dhellmann | where is lower-requirements.txt going to live and what's in it? | 15:39 |
| prometheanfire | lower-constraints.txt is what's tested | 15:39 |
| prometheanfire | lower-requirements.txt does not exist | 15:39 |
| prometheanfire | and should not exist really | 15:39 |
| dhellmann | "lower-requirements.txt can be dynamically generated from lower-constraints.txt" | 15:39 |
| prometheanfire | it can be, if desired, but it's not something we are looking to do | 15:39 |
| prometheanfire | you brought up that file name :P | 15:39 |
| dhellmann | did I? oh. | 15:40 |
| prometheanfire | think so, at least | 15:40 |
| prometheanfire | 09:36 < prometheanfire > that's why we need to store a lower-requirements.txt iirc (they become the project's lower bounds for stuff like that | 15:40 |
| prometheanfire | oh, maybe not | 15:40 |
| dhellmann | ok, I think we agree we can ignore "lower-requirements.txt" as a name | 15:40 |
| prometheanfire | yes | 15:40 |
| prometheanfire | we just need to have a lower-constraints file | 15:40 |
| dhellmann | we need the project requirements.txt, a project lower-constraints.txt, a global-requirements.txt, and a global upper-constraints.txt. | 15:40 |
| dhellmann | 4 files | 15:41 |
| dhellmann | 2 in openstack/requirements and 2 in openstack/$project | 15:41 |
| prometheanfire | yes | 15:41 |
| dirk | dhellmann: instead of having every project manage their own lower constarints, they could just inherit the global lower constraints (and maintain a "diff")? | 15:41 |
| prometheanfire | question | 15:41 |
| dhellmann | dirk : there cannot logically be a global lower-constraints file, because every project is going to (eventually) have different lower bounds in their requirements.txt files | 15:42 |
| prometheanfire | do we want to fully dynamically generate the lower-constraints.txt file | 15:42 |
| dhellmann | no, I think we need to track it in git just like we do for the global version, and for the same reason | 15:42 |
| dhellmann | if something goes wrong, we want to be able to manually change the file to fix the gate | 15:42 |
| dhellmann | because that file needs to list all of the second and third-order dependencies, just like upper-constraints.txt does | 15:43 |
| dhellmann | and we only need to generate it one time | 15:43 |
| prometheanfire | that's why I asked :P | 15:43 |
| dhellmann | after that, updating it is a manual process when someone updates the lower bound of a direct dependency | 15:43 |
| dhellmann | although we could probably provide a tool to make that simpler | 15:43 |
| prometheanfire | it will require a bootstrapping process for projects to start using it is all (a minor point) | 15:43 |
| dirk | dhellmann: well, there might be projects saying things like "want to be compatible with openstack series-1". so they could inherit lower-requirements from that series | 15:44 |
| dhellmann | yeah, we would need to add the tox environment that uses it and the job | 15:44 |
| prometheanfire | that tool can come later, better to focus on the simple stuff | 15:44 |
| dhellmann | dirk : how would you implement inheritance? | 15:44 |
| prometheanfire | anyway, I need to reboot again | 15:44 |
| dhellmann | prometheanfire : good luck! | 15:44 |
| dirk | dhellmann: using cat | sort -u | 15:45 |
| dirk | dhellmann: e.g. unless they overwrite it, they inherit | 15:45 |
| dhellmann | dirk : why not just do that one time and put the results in tree? | 15:45 |
| dirk | dhellmann: for tracking purposes (how do you know when you want to do an update what is by intention and what not?) | 15:46 |
| dhellmann | dirk : the lower bounds would only ever be updated intentionally. We would not have bots proposing those updates. | 15:46 |
| dhellmann | we need a tool to help us compute the settings one time, and that tool might be useful for updating the settings again later, but it would not be run automatically | 15:47 |
| dhellmann | the point is to lock those lower bounds to keep them from being raised unless that's needed | 15:47 |
| dhellmann | say when a new feature of a library is actually needed | 15:47 |
| dhellmann | at that point someone would update requirements.txt to set the new lower bound and they would also have to update the lower-constraints.txt file | 15:48 |
| prometheanfire | no luck needed, reboot was to re-enabled some sec options (age_poison=1 slab_nomerge vsyscall=none, slub_debug=P kills the zfs...) | 15:48 |
| dhellmann | oh, so not related to the zoom thing | 15:48 |
| prometheanfire | ya, I disabled those to see if they killed zoom | 15:48 |
| dhellmann | ah | 15:49 |
| prometheanfire | only other change was to disable pstates (maybe that's why my screen has this odd flicering (so says anothoer internet user)) | 15:49 |
| dirk | dhellmann: so a question, because otherwise we're going to discuss that endlessly: are you suggesting that every project should manage their openstack dependency versions independently, or only any 3rd party dependency? | 15:55 |
| openstackgerrit | Witold Bedyk proposed openstack/requirements master: Add monasca-tempest-plugin https://review.openstack.org/536873 | 15:55 |
| dhellmann | dirk : they should manage all of their *lower bound* dependencies independently and take the global upper constraints list. | 15:55 |
| *** coolsvap has quit IRC | 15:55 | |
| dhellmann | and that's all values in both sets, not just the openstack dependencies | 15:56 |
| dirk | dhellmann: how do they ensure that their local lower bound dependencies are meaningful when they use the global upper constraints? | 15:56 |
| dhellmann | there are 2 different types of tests. | 15:56 |
| dhellmann | well, 3 | 15:57 |
| dhellmann | 1. integration tests using the upper constraints list | 15:57 |
| dhellmann | 2. unit/functional tests using the upper constraints list | 15:57 |
| dhellmann | 3. unit/functional tests using the lower bounds list | 15:57 |
| dhellmann | that 3rd set of tests is only in-tree and is up to the project team to manage | 15:57 |
| dhellmann | the in-tree lower bounds need to be <= the upper-constraints list, but that's going to happen automatically because the tests in 1 and 2 will fail if it's not true. | 15:58 |
| dhellmann | we said in https://etherpad.openstack.org/p/ocata-requirements-notes that we don't need integration tests using the lower bounds because the purpose of supporting divergence there is for installing single components. if someone wants lower bounds with multiple components, they need to take steps to ensure isolation | 16:00 |
| *** openstackgerrit has quit IRC | 16:03 | |
| *** openstackgerrit has joined #openstack-requirements | 16:06 | |
| openstackgerrit | Dirk Mueller proposed openstack/requirements master: Add a global coinstallable lower-requirements set https://review.openstack.org/518416 | 16:06 |
| dirk | dhellmann: hmm, one more question (and I didn't read all of the backlog): are we going to stop tracking min versions in the global requirements repo alltogether, and every project has to individually manage it (also manage the transitive aspects)? | 16:07 |
| dhellmann | dirk : yes, that was the original proposal. if we let them diverge, there isn't a single global set to track. | 16:08 |
| dhellmann | we could make a tool that would let us combine all of the lower-constraints files from the individual projects, but that would give us a set of dependencies that had not been tested together. | 16:09 |
| dirk | well, that would the g-r/lower-constraints.txt (a combination of versions that has been passing the cross-* jobs/dsvm together) | 16:11 |
| dirk | dhellmann: do you expect that all openstack projects start trakcing their own lower req individually or do you think it will be the 1-2 projects only? | 16:12 |
| dirk | (e.g. swift or whatever is the other typical example) | 16:12 |
| dirk | I would assume that many projects are going to depend on other projects within openstack still | 16:12 |
| dhellmann | I think we'll do it for the oslo libraries, too | 16:12 |
| dhellmann | I would work on implementing that, for example | 16:12 |
| dhellmann | some of the other projects that emphasize their ability to be installed and used on their own might, too | 16:13 |
| dhellmann | cinder, for example | 16:13 |
| dhellmann | although I haven't asked any of them specifically | 16:13 |
| dirk | dhellmann: hmm, I'll need to think about it. that basically means we shouldn"t worry about a coinstallable global lower constraints. I'd still not want to see that going given that e..g devstack has a much higher test coverage than unit tests/functional tests of all the projects, but I understand the general trend is to do indivdal service containers rather than a sensible set of common versions (that avoid distros having to maintain the | 16:26 |
| dirk | same dependency in 5 different versions) | 16:26 |
| dhellmann | dirk : right, having co-installable lower-constraints was not a requirement I considered originally. in-tree functional tests would take the place of those integration tests for adding coverage for the lower-bounds because the point of diverging is to let the component be used on its own | 16:28 |
| dhellmann | and our full-cloud integration tests are less interesting for that use case | 16:28 |
| dhellmann | dirk, prometheanfire : I also suggest we only test lower constraints using python 3 | 16:29 |
| dhellmann | since the versions for 3 might be higher than for 2 anyway | 16:29 |
| dhellmann | although if a project wanted to support testing 2 and 3 I wouldn't object to them doing the work to set that up | 16:29 |
| dirk | dhellmann: makes sense | 16:31 |
| dhellmann | I'll try to summarize our discussion from today in an action plan of some sort before the ptg | 16:32 |
| prometheanfire | dhellmann: I'm fine with droping 2.7 globally :P | 16:36 |
| prometheanfire | when is 2.7 eol again? | 16:36 |
| prometheanfire | lol https://pythonclock.org/ | 16:37 |
| dhellmann | we should make an ICS file with weekly countdown information | 16:50 |
| prometheanfire | that'd be nice | 16:52 |
| openstackgerrit | Merged openstack/requirements master: update constraint for neutron-lib to new release 1.13.0 https://review.openstack.org/536627 | 17:00 |
| openstackgerrit | Merged openstack/requirements master: update constraint for os-brick to new release 2.3.0 https://review.openstack.org/535526 | 17:00 |
| *** cjloader has quit IRC | 17:20 | |
| *** cjloader has joined #openstack-requirements | 17:21 | |
| *** jpich has quit IRC | 17:39 | |
| openstackgerrit | Merged openstack/requirements master: show values at -eol tags as well as on branches https://review.openstack.org/536606 | 17:58 |
| *** ralonsoh__ has quit IRC | 18:03 | |
| *** cjloader has quit IRC | 19:00 | |
| *** cjloader has joined #openstack-requirements | 19:00 | |
| openstackgerrit | Brian Rosmaita proposed openstack/requirements master: Revert "disable glance voting on py27 tests" https://review.openstack.org/536946 | 19:53 |
| openstackgerrit | Brian Rosmaita proposed openstack/requirements master: Revert "disable glance voting on py27 tests" https://review.openstack.org/536946 | 19:54 |
| prometheanfire | smcginnis: thanks :D | 20:01 |
| prometheanfire | smcginnis: what was the problem? | 20:01 |
| smcginnis | prometheanfire: With glance? There's still a problem, but we've now split out the unit tests and the functional tests so they don't all run in one job. | 20:07 |
| smcginnis | prometheanfire: So for the sake of some coverage, we'd like to get the unit tests back voting. In the meantime, we're trying to figure out what is causing the functional test failures. | 20:08 |
| *** edmondsw has quit IRC | 20:08 | |
| smcginnis | Appears to be a system package update that is causing some failure internally. I don't think anyone has tracked down where the failure is exactly. Nor which package is the cause. | 20:09 |
| *** edmondsw has joined #openstack-requirements | 20:09 | |
| prometheanfire | smcginnis: ah, k | 20:09 |
| smcginnis | I think someone mentioned seeing some possible rootwrap issues, so if you happen to know of any Ubuntu package updates that could impact rootwrap, I'm sure they would love to know about it. :) | 20:11 |
| prometheanfire | nope, just gentoo :P | 20:11 |
| *** edmondsw has quit IRC | 20:13 | |
| *** edmondsw has joined #openstack-requirements | 20:15 | |
| *** edmondsw has quit IRC | 20:19 | |
| *** Jeffrey4l has quit IRC | 20:25 | |
| *** Jeffrey4l has joined #openstack-requirements | 20:25 | |
| *** edmondsw has joined #openstack-requirements | 20:26 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: update constraint for python-troveclient to new release 2.14.0 https://review.openstack.org/536963 | 20:50 |
| openstackgerrit | boden proposed openstack/requirements master: bump requirements for neutron-lib to 1.13.0 https://review.openstack.org/536964 | 20:50 |
| *** edmondsw has quit IRC | 21:33 | |
| openstackgerrit | Merged openstack/requirements master: Raise minimum version of openstackdocstheme to 1.18.1 https://review.openstack.org/536670 | 21:33 |
| *** edmondsw has joined #openstack-requirements | 21:33 | |
| *** edmondsw has quit IRC | 21:38 | |
| *** cjloader has quit IRC | 23:05 | |
| *** oanson has quit IRC | 23:42 | |
| *** oanson has joined #openstack-requirements | 23:43 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!