| *** KeithMnemonic has quit IRC | 04:30 | |
| *** evrardjp has quit IRC | 04:37 | |
| *** evrardjp has joined #openstack-requirements | 04:37 | |
| *** udesale has joined #openstack-requirements | 05:37 | |
| *** vishalmanchanda has joined #openstack-requirements | 05:40 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/requirements master: Updated from generate-constraints https://review.opendev.org/719345 | 06:20 |
|---|---|---|
| *** ccamacho has joined #openstack-requirements | 07:11 | |
| *** ralonsoh has joined #openstack-requirements | 07:14 | |
| *** tosky has joined #openstack-requirements | 07:23 | |
| *** rpittau|afk is now known as rpittau | 07:28 | |
| *** iurygregory has quit IRC | 07:31 | |
| *** iurygregory has joined #openstack-requirements | 07:32 | |
| *** e0ne has joined #openstack-requirements | 08:12 | |
| *** ccamacho has quit IRC | 08:20 | |
| *** ccamacho has joined #openstack-requirements | 08:40 | |
| *** ccamacho has quit IRC | 08:40 | |
| *** ccamacho has joined #openstack-requirements | 08:45 | |
| *** dtantsur|afk is now known as dtantsur | 08:49 | |
| *** ccamacho has quit IRC | 09:44 | |
| *** ccamacho has joined #openstack-requirements | 10:05 | |
| *** rpittau is now known as rpittau|bbl | 10:23 | |
| *** e0ne has quit IRC | 10:52 | |
| *** e0ne has joined #openstack-requirements | 10:53 | |
| *** iurygregory is now known as iurygregory_afk | 11:31 | |
| *** udesale_ has joined #openstack-requirements | 12:04 | |
| *** udesale has quit IRC | 12:06 | |
| smcginnis | AJaeger: The reason why the 3.8 line wasn't updated is the project doesn't include 3.8 in its metadata. | 12:13 |
| smcginnis | So the script extracts the declared Python versions from the metadata and only matches up the lines that are the same versions. | 12:14 |
| *** brinzhang_ has quit IRC | 12:14 | |
| smcginnis | That was to be able to handle 2 vs 3, but also to be able to handle the case when we get to the point of dropping 3.6 but supporting newer versions. | 12:14 |
| smcginnis | Definitely not perfect. I suppose we could just assume any py3 version should match. Wouldn't be too hard to update the sed statement that does that. | 12:15 |
| AJaeger | smcginnis: ah, interesting. With us going to 3.8, I think we should change it somehow - especially since the constrain generator updates it just. So, that should be consistent... | 12:29 |
| smcginnis | Yeah. Only updating if the project declared 3.8 support seems more "correct", but we're going to have some transition time until that happens. | 12:33 |
| smcginnis | I may not be able to look at it for a bit, but if you are curious, here is the code that handles that: https://review.opendev.org/#/c/709809/2/roles/copy-release-tools-scripts/files/release-tools/update_constraints.sh | 12:34 |
| smcginnis | Should be reasonable to just check if $PYTHON_3_VERSIONS are present, and if so, just do one update for anything "3.*" | 12:35 |
| *** rpittau|bbl is now known as rpittau | 12:49 | |
| *** KeithMnemonic has joined #openstack-requirements | 12:54 | |
| AJaeger | smcginnis, prometheanfire, what do you think of https://review.opendev.org/720197 ? | 12:58 |
| KeithMnemonic | smcginnis, prometheanfire i think this patch broke monasca-agent , https://review.opendev.org/#/c/717960/ the error in zuul shows http://paste.openstack.org/show/792161/ | 13:23 |
| KeithMnemonic | do you know how i can fix this ? | 13:23 |
| smcginnis | KeithMnemonic: 1.4.0 was released on the 9th: https://pypi.org/project/confluent-kafka/#history | 13:25 |
| smcginnis | KeithMnemonic: Maybe a stale pypi mirror? Did you try a recheck? | 13:25 |
| KeithMnemonic | and that is when monasca-agent started failing | 13:26 |
| KeithMnemonic | yes 3 times yesterday | 13:26 |
| KeithMnemonic | https://zuul.opendev.org/t/openstack/builds?job_name=build-monasca-docker-image&branch=master | 13:27 |
| KeithMnemonic | last good run was 4/8 | 13:27 |
| KeithMnemonic | so right before that patch landed | 13:27 |
| smcginnis | KeithMnemonic: I was able to do "pip install confluent-kafka==='1.4.0'" locally, so something else is going on there. | 13:27 |
| AJaeger | KeithMnemonic: so, if that is not able to be installed, then let's discuss on #opendev... | 13:32 |
| KeithMnemonic | intersting there are two failure on my patch. the other one on seems to be able to install it "Successfully installed PyMySQL-0.9.3 confluent-kafka-1.4.0" | 13:35 |
| KeithMnemonic | that is from "monasca-tempest-python3-influxdb " | 13:35 |
| KeithMnemonic | so it seems maybe just the docker image has issues? | 13:36 |
| KeithMnemonic | with kafka? | 13:36 |
| smcginnis | Really seems like maybe an infra thing. The package is there matching the upper-constraint, so nothing from that perspective to indicate why installation failed. | 13:36 |
| smcginnis | Anything different in how the docker image is created? | 13:36 |
| AJaeger | smcginnis: looks like they pushed only wheels, see https://pypi.org/project/confluent-kafka/#files | 13:37 |
| AJaeger | KeithMnemonic: ^ | 13:37 |
| KeithMnemonic | ok so who can fix that? | 13:38 |
| AJaeger | so, first question: IS that the problem? If it is: Only the confluent-kafka authors | 13:38 |
| KeithMnemonic | https://71c0eda8b1d9d45ff8c8-9e1e29a469f8f669289667ecce245a85.ssl.cf5.rackcdn.com/718195/3/check/build-monasca-docker-image/67cd43f/ | 13:39 |
| KeithMnemonic | that is the log | 13:39 |
| smcginnis | https://github.com/confluentinc/confluent-kafka-python/issues/830 | 13:39 |
| AJaeger | smcginnis: so, back to confluent-kafka 1.3.0 for now? | 13:40 |
| AJaeger | KeithMnemonic: want to ask for FFE and sent a patch referencing the github issue? | 13:40 |
| smcginnis | The two consumers of it are monasca-common and oslo.messaging. | 13:41 |
| smcginnis | oslo.messaging doesn't appear to be broken. | 13:41 |
| smcginnis | I wonder if there's something that can be changed in how monasca-common is trying to install this? | 13:41 |
| KeithMnemonic | does oslo.messaging use the wheel | 13:41 |
| KeithMnemonic | the last patch that passed shows it pulls the tar | 13:42 |
| AJaeger | smcginnis: I rechecked https://review.opendev.org/#/c/717600 - let's see whether it fails | 13:42 |
| KeithMnemonic | last patch for monasca agent i mean | 13:42 |
| KeithMnemonic | it pulls the wheel for oslo.messaging | 13:44 |
| AJaeger | interesting that they do it differently | 13:45 |
| smcginnis | KeithMnemonic: Not directly related, but there's a bug in those docker builds that they are not following stable branch constraints: https://opendev.org/openstack/monasca-agent/src/branch/stable/train/docker/forwarder/build_image.sh#L67 | 13:47 |
| smcginnis | Oh, maybe they are later on in the file. | 13:48 |
| KeithMnemonic | so i guess i either get the monasca guys to refactor to use the wheel or wait for the confluent issue to get resolved | 14:23 |
| AJaeger | KeithMnemonic: or go back to previous version in constraints | 14:25 |
| KeithMnemonic | can i put a patch up for that smcginnis: ? | 14:26 |
| smcginnis | KeithMnemonic: I think updating monasca to work with that is the best option. Otherwise contact the confluent-kafka team to ask that they publish more than just the wheel. | 14:27 |
| smcginnis | KeithMnemonic: But given it's limited use, we could probably lower that constraint for now. The problem being unless we completely blacklist that version, the nightly constraints update job is just going to propose raising it again. | 14:28 |
| KeithMnemonic | thanks, i brought this open in the monasca channel as well. so far crickets but lets see if they can address this in a patch | 14:30 |
| KeithMnemonic | i am not sure i know enough how to change it from tar to whl | 14:30 |
| *** witek has joined #openstack-requirements | 14:32 | |
| smcginnis | Yeah, I'm not sure how things are being done there, and unfortunately I can't dig into it right now to be able to offer any suggestions. Hoping someone on that team understands it enough that the fix ends up being fairly obvious. | 14:33 |
| witek | the Docker job uses Alpine and confluent-kafka doesn't build wheels for that, from what I recall | 14:34 |
| *** dtantsur is now known as dtantsur|brb | 14:55 | |
| KeithMnemonic | they fixed it upstream, there is a tar now | 15:32 |
| KeithMnemonic | that was quick | 15:32 |
| AJaeger | great, so want to recheck? | 15:35 |
| prometheanfire | AJaeger: smcginnis did more of that sed work, he'd be the person to run the https://review.opendev.org/720197 by | 15:41 |
| *** iurygregory_afk is now known as iurygregory | 15:41 | |
| prometheanfire | AJaeger: https://opendev.org/openstack/project-config/commit/45ed2e8fa83b6673e69586e735356db697349cef may be the problem as well | 15:43 |
| AJaeger | prometheanfire: yes, I asked smcginnis for 720197 review | 15:44 |
| prometheanfire | AJaeger: what I'd like to do is change the generate constraints call to be python3.6/3.8 only instead of py2.7/3.6 with coppying 3.6 to 3.8 | 15:45 |
| smcginnis | I think the propose_update.sh is fine. It was just a matter that we've only been updating new releases based on what they declare in their metadata, which pretty much everyone does not include 3.8 yet. | 15:45 |
| AJaeger | thx for review | 15:45 |
| AJaeger | smcginnis: will there be an automatic add for 3.8 to setup.cfg? | 15:45 |
| smcginnis | AJaeger: No. :/ | 15:47 |
| smcginnis | Not sure if we could do that, but it would be good to add that to the new series job template update automation. | 15:47 |
| AJaeger | yeah, would be nice | 15:47 |
| smcginnis | 3.9 is out by the end of this year, but I would guess after we update for 3.8, it will probably be some time before distros will need us to add 3.9. | 15:48 |
| prometheanfire | ya, 3.8 isn't even fully stable here yet | 15:49 |
| prometheanfire | seeing what happens when I run this locally generate-constraints -b blacklist.txt -p /usr/bin/python3.6 -p /usr/bin/python3.8 -r global-requirements.txt | 16:05 |
| prometheanfire | that should give us 'true' upper-constraints I think. | 16:06 |
| *** dtantsur|brb is now known as dtantsur | 16:11 | |
| prometheanfire | worked, kinda, salt uses platform_distribution, which is gone in py38 | 16:18 |
| *** witek has quit IRC | 16:23 | |
| *** rpittau is now known as rpittau|afk | 16:33 | |
| *** evrardjp has quit IRC | 16:37 | |
| *** evrardjp has joined #openstack-requirements | 16:37 | |
| *** udesale_ has quit IRC | 16:49 | |
| *** dtantsur is now known as dtantsur|afk | 16:53 | |
| *** ralonsoh has quit IRC | 17:55 | |
| prometheanfire | smcginnis: I had to mask salt as it doesn't work with py38 yet, but this is the result I get if I run with the above command https://gist.github.com/38d526b487513326467356ab2d659d0e | 18:21 |
| prometheanfire | not bad, imo | 18:22 |
| smcginnis | Great! | 18:22 |
| prometheanfire | yep, as soon as we branch | 18:23 |
| *** e0ne has quit IRC | 18:31 | |
| openstackgerrit | Matthew Thode proposed openstack/requirements master: DNM: test py3 only constraints for victoria https://review.opendev.org/720264 | 18:47 |
| prometheanfire | upper-constraints.txt +149, -568 | 18:48 |
| prometheanfire | heh | 18:48 |
| smcginnis | I love reducing lines of code. :) | 18:56 |
| prometheanfire | same | 18:57 |
| prometheanfire | a great culling is upon us | 18:57 |
| prometheanfire | there's some special code around py27 as well I think | 18:57 |
| * prometheanfire should remove lines from gr as well | 18:57 | |
| *** hongbin has joined #openstack-requirements | 19:01 | |
| smcginnis | I've thought it would be good to do an audit of gr to see how many entries are no longer used. | 19:09 |
| prometheanfire | ya, that could be a port of the patch, just not automated :D | 19:13 |
| prometheanfire | atm, doing cross gates with the new constraints is a start | 19:13 |
| smcginnis | ++ | 19:17 |
| prometheanfire | quite a few failures | 19:19 |
| prometheanfire | 2020-04-15 18:54:10.343849 | ubuntu-bionic | 'contextlib2' appears in global-requirements.txt but not upper-constraints.txt or blacklist.txt | 19:20 |
| prometheanfire | gonna be a few of those | 19:20 |
| smcginnis | Ah. So maybe we do need to do some gr cleanup along with this. | 19:24 |
| smcginnis | Or add a bunch of stuff to uc that we don't necessarily need. | 19:24 |
| openstackgerrit | Matthew Thode proposed openstack/requirements master: DNM: test py3 only constraints for victoria https://review.opendev.org/720264 | 19:33 |
| prometheanfire | smcginnis: yep | 19:33 |
| prometheanfire | +217, -713 | 19:33 |
| prometheanfire | only two lines specifying python versions now | 19:34 |
| prometheanfire | 34:dataclasses;python_version=='3.6' # Apache-2.0 | 19:34 |
| prometheanfire | 226:salt!=2019.2.1,!=2019.2.2;python_version!='3.8' # Apache-2.0 | 19:34 |
| prometheanfire | was able to remove things like this too :D | 19:35 |
| prometheanfire | -monotonic;python_version<'3.3' # Apache-2.0 | 19:35 |
| AJaeger | prometheanfire: nice! | 19:36 |
| prometheanfire | AJaeger: I want it | 19:36 |
| prometheanfire | have zuul updates to go into the next ammended commit | 19:43 |
| prometheanfire | removing py37 and py27 (that means swift) | 19:46 |
| *** vishalmanchanda has quit IRC | 19:54 | |
| timburke | oh yeah, i should take over https://review.opendev.org/#/c/712984/ -- i started to sort through deps a little, figure out what's already done their final py2 release, and what the most likely sources of breakage will be (things like iso8601 which haven't been updated in years seem unlikely to ever be updated again at this point) | 19:54 |
| prometheanfire | timburke: :D | 19:56 |
| * prometheanfire needs to look at the swift configparser bug | 19:57 | |
| timburke | it's a mess -- i've got a dirty tree to try to sort it out -- i was hoping i might be able to get away with not monkey-patching like it did in the original patch, but idk... | 19:59 |
| prometheanfire | 'one right way of doing things' | 20:00 |
| prometheanfire | timburke: I do have a question about that though, you said to just put it in one config section as the config is only used in that part of swift right? | 20:00 |
| timburke | no, i said that but i was wrong. *maybe* you could get away with scattering it between a bunch of different sections and having some of them say 1% and others 1%% -- but ugh | 20:03 |
| prometheanfire | timburke: yep | 20:03 |
| timburke | is it configurable in OSA? or is it hardcoded to 1%? because that should be the default in swift; you might consider just removing the option from the configs that OSA's pushing out... | 20:03 |
| prometheanfire | timburke: atm I have two versions of the config, depending on the service that's running (see https://review.opendev.org/720021) | 20:04 |
| prometheanfire | it's configurable, but it's either 1% or 1%% | 20:04 |
| prometheanfire | my review moves the % (or %%) character(s) to the template from the config | 20:04 |
| prometheanfire | s/config/defaults | 20:05 |
| prometheanfire | timburke: I don't think applying it to all those sections will work, configparser will probably error out still as it still has to parse the whole file | 20:16 |
| openstackgerrit | Matthew Thode proposed openstack/requirements master: DNM: test py3 only constraints for victoria https://review.opendev.org/720264 | 20:16 |
| *** e0ne has joined #openstack-requirements | 20:24 | |
| prometheanfire | !startmeeting requirements | 20:30 |
| openstack | prometheanfire: Error: "startmeeting" is not a valid command. | 20:30 |
| prometheanfire | #startmeeting requirements | 20:30 |
| openstack | Meeting started Wed Apr 15 20:30:05 2020 UTC and is due to finish in 60 minutes. The chair is prometheanfire. Information about MeetBot at http://wiki.debian.org/MeetBot. | 20:30 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 20:30 |
| *** openstack changes topic to " (Meeting topic: requirements)" | 20:30 | |
| openstack | The meeting name has been set to 'requirements' | 20:30 |
| prometheanfire | `#topic rollcall | 20:30 |
| prometheanfire | tonyb, prometheanfire, dirk, smcginnis ping | 20:30 |
| prometheanfire | #topic rollcall | 20:30 |
| *** openstack changes topic to "rollcall (Meeting topic: requirements)" | 20:30 | |
| prometheanfire | tonyb, prometheanfire, dirk, smcginnis ping | 20:30 |
| smcginnis | o/ | 20:30 |
| prometheanfire | o/ | 20:30 |
| prometheanfire | #topic Any controversies in the Queue? | 20:31 |
| *** openstack changes topic to "Any controversies in the Queue? (Meeting topic: requirements)" | 20:31 | |
| prometheanfire | we are frozen, so I hope not :D | 20:31 |
| prometheanfire | https://zuul.openstack.org/status/change/720264,3 seems to be going well neutron/nova failed last run, but I think that's because it had stuff like updated msgpack causing issues | 20:32 |
| prometheanfire | #topic open floor | 20:33 |
| *** openstack changes topic to "open floor (Meeting topic: requirements)" | 20:33 | |
| prometheanfire | ya, nova failed :| | 20:33 |
| smcginnis | There was something else with msgpack I thought. | 20:33 |
| smcginnis | Nah, I think I'm thinking of something else. | 20:34 |
| prometheanfire | something caps it so we can't use the update | 20:34 |
| prometheanfire | https://d77580f8adad064c9e13-1cbf7c2bad1b53c710605a1cfc31790e.ssl.cf5.rackcdn.com/720264/3/check/cross-nova-py36/d516452/testr_results.html for some light reading | 20:34 |
| smcginnis | Fun | 20:34 |
| prometheanfire | ya, will pass it by the nova team later | 20:35 |
| prometheanfire | anything else before I close this pro-forma meeting :D | 20:35 |
| smcginnis | Nothing else from me. I think we're on good track for ussuri. | 20:35 |
| prometheanfire | same | 20:35 |
| smcginnis | Victoria should be a lot of fun. :D | 20:35 |
| prometheanfire | #endmeeting | 20:35 |
| *** openstack changes topic to "OpenStack Requirements - IRC meetngs on Wednesdays @ 07:00 UTC in here in #openstack-requirements - See agenda @ http://tinyurl.com/h44ryuw - IRC channel is *LOGGED* @ http://tinyurl.com/j38rk24" | 20:35 | |
| openstack | Meeting ended Wed Apr 15 20:35:36 2020 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 20:35 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/requirements/2020/requirements.2020-04-15-20.30.html | 20:35 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/requirements/2020/requirements.2020-04-15-20.30.txt | 20:35 |
| prometheanfire | indeed, a bunch of cleanup :D | 20:35 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/requirements/2020/requirements.2020-04-15-20.30.log.html | 20:35 |
| smcginnis | timburke: Does swift still plan on supporting py27 in the victoria cycle? | 20:36 |
| prometheanfire | oh, those errors are from the updated mock | 20:36 |
| prometheanfire | that's right | 20:36 |
| openstackgerrit | Matthew Thode proposed openstack/requirements master: DNM: test py3 only constraints for victoria https://review.opendev.org/720264 | 20:37 |
| timburke | smcginnis, yes, that's the current plan | 20:39 |
| smcginnis | Updated mock? Like the third party one? | 20:39 |
| smcginnis | timburke: OK, great. Was just curious, but glad to hear. Should simplify some things going forward. | 20:39 |
| timburke | i was expecting more pushback ;-) | 20:40 |
| smcginnis | timburke: Hah, I just realized I misread that. Though you answered what I asked, I was thinking you said the plan was to drop py27. :) | 20:41 |
| smcginnis | Oh well. That should be fairly static now. | 20:41 |
| timburke | that's part of my thinking -- most of what breaks will break now, after which point, may as well continue | 20:43 |
| timburke | just this morning we had someone in irc running swift 2.7.0 (so, mitaka) on xenial and running into troubles. i could in good conscience recommend an upgrade to at least 2.22.0, and the only reason not to go with latest release was an update to our minimum version of eventlet | 20:47 |
| prometheanfire | smcginnis: ya, the change to built-in mock is something for victoria as well | 20:47 |
| smcginnis | timburke: Well, for 12 more months until xenial itself is EOL. ;) | 20:48 |
| timburke | just because it's EOL doesn't mean it goes away :P | 20:48 |
| smcginnis | prometheanfire: I've been trying to get as many as I can going already this cycle. | 20:48 |
| smcginnis | timburke: No, but it gives a good reason to recommend someone takes the effort to actually upgrade their deployment. | 20:49 |
| *** e0ne has quit IRC | 21:02 | |
| *** e0ne has joined #openstack-requirements | 21:03 | |
| prometheanfire | just functests to go | 21:16 |
| *** e0ne has quit IRC | 21:42 | |
| *** ccamacho has quit IRC | 22:11 | |
| *** prometheanfire has quit IRC | 22:17 | |
| *** prometheanfire has joined #openstack-requirements | 22:39 | |
| prometheanfire | and:D | 22:41 |
| *** hongbin has quit IRC | 22:53 | |
| *** hongbin has joined #openstack-requirements | 23:01 | |
| *** tosky has quit IRC | 23:39 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!