| fungi | are we also removing the jvb service from meetpad01 so the other services don't have to compete for resources? | 00:04 |
|---|---|---|
| clarkb | fungi: no | 00:04 |
| clarkb | fungi: when we looked at it during the openstack release celebration the jvb is basically the only resource consumer | 00:04 |
| ianw | fungi: did you look into what the afs mount path should be for buster wheels? | 00:04 |
| clarkb | I think it will be ok to run it alongside the other services, but if it isn't then we can split things up further and deploy a smaller meetpad | 00:04 |
| clarkb | though thats all likely post ptg I think | 00:05 |
| *** diablo_rojo has quit IRC | 00:05 | |
| clarkb | hrm cacti hasn't managed to pull data from meetpad | 00:08 |
| clarkb | I've restarted snmpd on meetpad just to rule that out but haven't seen any change over what I think has been a couple collection periods | 00:08 |
| clarkb | maybe I do need to wait a bit longer, but I've also checked the firewall rules and it looks happy there too | 00:08 |
| fungi | ianw: i hadn't gotten that far yet, nope, sorry | 00:16 |
| ianw | that's ok, it's going to be "debian-10-x86_64" | 00:19 |
| fungi | looks right | 00:20 |
| openstackgerrit | Ian Wienand proposed openstack/project-config master: Match Debian in AFS slug generation https://review.opendev.org/731009 | 00:20 |
| ianw | fungi: after something like ^ | 00:20 |
| ianw | fungi: cna you confirm you can see everything in /afs/openstack.org/mirror/wheel now? | 00:22 |
| fungi | checking | 00:24 |
| ianw | hrm i get forbidden on https://mirror.iad.rax.opendev.org/wheel/ubuntu-20.04-x86_64/ ... acl's should be right | 00:27 |
| fungi | ianw: yeah, looks good for 730323, want to do the ones for 730342 as well (going with "a64" as the suffix)? | 00:27 |
| markmcclain | fungi/clarkb: sorry... toddlers pull me away for a bit. I'll check to see if there was an out of band manual upload | 00:27 |
| ianw | fungi: yeah i can ... have to think about the slug and the mount point there again too | 00:28 |
| fungi | markmcclain: looks like some manual step has been performed by the openstack-arista account for every release from april to now, which coincides with the upload post_failure results | 00:28 |
| ianw | i think ansible arch is "aarch64" | 00:28 |
| fungi | ianw: yeah, but we need the volume names to fit in 22 bytes | 00:28 |
| fungi | (see discussion in 730342) | 00:29 |
| clarkb | ianw: fungi shorten to arm? | 00:29 |
| fungi | we could, i guess | 00:30 |
| fungi | we could also shorten wheel to whl... there are a number of possibilities | 00:30 |
| ianw | fungi: i'm thinking a64 (as opposed to x64) is probably best for consistency | 00:34 |
| clarkb | ++ | 00:35 |
| ianw | i get that it's confusing, but every choice is at this point, i don't really feel like renaming all the extant ones | 00:35 |
| ianw | as you say, we mount it at a more descriptive name | 00:35 |
| ianw | ok, i need a cup of tea but will do that :) | 00:36 |
| *** Meiyan has joined #opendev | 00:56 | |
| openstackgerrit | Merged openstack/project-config master: Update pip output parsing to fix wheel mirror builds https://review.opendev.org/730933 | 01:03 |
| openstackgerrit | Merged openstack/project-config master: Update neutron stable grafana dashboards https://review.opendev.org/729291 | 01:04 |
| markmcclain | fungi: Thanks for helping dig into this. Found out there was an out-of-date internal doc that steering our internal teammates to add the conflicting manual upload. | 01:39 |
| fungi | markmcclain: glad we could sort that. if you get a chance to tag another release without doing that step it might be good to double-check the plumbing is still working | 01:40 |
| fungi | but at least the error makes more sense now | 01:41 |
| *** factor has quit IRC | 01:56 | |
| ianw | fungi: ok, a little cross-eyed and lots of fs checkvolumes later but i think http://mirror.iad.rax.opendev.org/wheel/ has the volumes required | 02:53 |
| *** ysandeep|away is now known as ysandeep | 03:38 | |
| openstackgerrit | Ian Wienand proposed openstack/project-config master: AFS grafana : add wheel volumes https://review.opendev.org/731072 | 04:25 |
| openstackgerrit | Ian Wienand proposed openstack/project-config master: build and publish wheels for aarch64 architecture https://review.opendev.org/730342 | 04:31 |
| openstackgerrit | Merged openstack/diskimage-builder master: dib-lint: use yamllint to parse YAML files https://review.opendev.org/730690 | 04:48 |
| openstackgerrit | Merged openstack/diskimage-builder master: package-installs : allow a list of parameters https://review.opendev.org/730691 | 04:48 |
| openstackgerrit | Merged openstack/diskimage-builder master: Revert "Revert "ubuntu-minimal : only install 16.04 HWE kernel on xenial"" https://review.opendev.org/730692 | 04:50 |
| openstackgerrit | Merged opendev/system-config master: Use ipv4 in server launch inventory output https://review.opendev.org/730149 | 05:13 |
| *** ykarel|away is now known as ykarel | 05:44 | |
| *** DSpider has joined #opendev | 05:44 | |
| *** hashar has joined #opendev | 05:57 | |
| *** dpawlik has joined #opendev | 06:05 | |
| *** dpawlik has quit IRC | 06:12 | |
| *** dpawlik has joined #opendev | 06:12 | |
| *** sgw has quit IRC | 06:50 | |
| openstackgerrit | Anand Shanmugam proposed openstack/project-config master: Remove py27 and pypi publish job from cloudpulse https://review.opendev.org/731089 | 07:17 |
| openstackgerrit | Anand Shanmugam proposed openstack/project-config master: Remove py27 and pypi publish job from cloudpulse https://review.opendev.org/731089 | 07:19 |
| *** tosky has joined #opendev | 07:26 | |
| *** slaweq has joined #opendev | 07:32 | |
| *** bolg has joined #opendev | 07:56 | |
| *** bolg has quit IRC | 07:59 | |
| *** mlavalle has quit IRC | 08:00 | |
| *** mlavalle has joined #opendev | 08:00 | |
| *** bolg has joined #opendev | 08:00 | |
| *** moppy has quit IRC | 08:01 | |
| *** moppy has joined #opendev | 08:01 | |
| *** mnaser_ has joined #opendev | 08:13 | |
| *** mugsie_ has joined #opendev | 08:14 | |
| *** mnaser has quit IRC | 08:20 | |
| *** melwitt has quit IRC | 08:20 | |
| *** mugsie has quit IRC | 08:20 | |
| *** cmurphy has quit IRC | 08:20 | |
| *** mnaser_ is now known as mnaser | 08:20 | |
| *** rpittau|afk is now known as rpittau | 08:22 | |
| *** _mlavalle_1 has joined #opendev | 08:22 | |
| *** slaweq has quit IRC | 08:22 | |
| *** mlavalle has quit IRC | 08:24 | |
| *** tkajinam has quit IRC | 08:27 | |
| hrw | clarkb, fungi, ianw: thanks! | 08:27 |
| *** dtantsur|afk is now known as dtantsur | 08:32 | |
| openstackgerrit | Anand Shanmugam proposed openstack/project-config master: Remove gate/check jobs for cloudpulse and cloudpulseclient https://review.opendev.org/731089 | 08:41 |
| *** ysandeep is now known as ysandeep|lunch | 08:46 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add ensure-dnf-copr https://review.opendev.org/730743 | 08:47 |
| *** slaweq has joined #opendev | 08:51 | |
| *** slaweq has quit IRC | 08:56 | |
| *** slaweq has joined #opendev | 09:20 | |
| openstackgerrit | Merged opendev/irc-meetings master: Add Secure Default policies popup team meeting https://review.opendev.org/730935 | 09:25 |
| *** sshnaidm|afk is now known as sshnaidm | 09:26 | |
| *** ykarel is now known as ykarel|lunch | 09:35 | |
| *** ysandeep|lunch is now known as ysandeep | 09:39 | |
| *** lpetrut has joined #opendev | 10:02 | |
| *** rpittau is now known as rpittau|bbl | 10:07 | |
| *** slaweq has quit IRC | 10:25 | |
| *** roman_g has joined #opendev | 10:34 | |
| *** ykarel|lunch is now known as ykarel | 10:36 | |
| *** hashar has quit IRC | 11:16 | |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 11:21 |
| *** sshnaidm is now known as sshnaidm|afk | 11:25 | |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 11:27 |
| *** Meiyan has quit IRC | 11:29 | |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 11:34 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: upload-artifactory: documentation fix https://review.opendev.org/731170 | 11:36 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: fetch-subunit-output: stop using system os-testr https://review.opendev.org/730482 | 11:43 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: ensure-haskell-stack: Make sure dnf copr is installed https://review.opendev.org/731173 | 11:46 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 11:50 |
| openstackgerrit | Sorin Sbarnea (zbr) proposed zuul/zuul-jobs master: Bump ansible-lint to 4.3.0 https://review.opendev.org/702679 | 11:57 |
| *** ykarel is now known as ykarel|afk | 11:59 | |
| *** rpittau|bbl is now known as rpittau | 12:06 | |
| *** elod_ has joined #opendev | 12:08 | |
| *** elod has quit IRC | 12:09 | |
| *** elod_ is now known as elod | 12:12 | |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 12:17 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: tox: envlist bugfixes https://review.opendev.org/730381 | 12:32 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: siblings: python2.7 configparser has no __getitem__ https://review.opendev.org/731187 | 12:37 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: siblings: python2.7 ConfigParser has no __getitem__ https://review.opendev.org/731187 | 12:38 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: siblings: python2.7 ConfigParser has no __getitem__ https://review.opendev.org/731187 | 12:39 |
| *** ykarel|afk is now known as ykarel | 12:41 | |
| *** sshnaidm|afk is now known as sshnaidm | 12:45 | |
| *** cmurphy has joined #opendev | 13:00 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: siblings: python2.7 ConfigParser has no __getitem__ https://review.opendev.org/731187 | 13:04 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Constrain soupsieve to <2 for python2.7 https://review.opendev.org/731195 | 13:04 |
| *** sgw has joined #opendev | 13:05 | |
| *** lpetrut has quit IRC | 13:18 | |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Move base roles into a base subdir https://review.opendev.org/730937 | 13:41 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Split out a base playbook for the zuul service https://review.opendev.org/730999 | 13:41 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Split inventory into multiple dirs and move hostvars https://review.opendev.org/730991 | 13:41 |
| mordred | corvus: I had a thought in the middle of the night that the hostvar split idea wouldn't work right for $reason - but I don't remember what that reason was | 13:41 |
| corvus | mordred: were you awake or asleep when you had the thought? :) | 13:42 |
| mordred | corvus: yes | 13:42 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Constrain soupsieve to <2 for python2.7 https://review.opendev.org/731195 | 13:43 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: siblings: python2.7 ConfigParser has no __getitem__ https://review.opendev.org/731187 | 13:43 |
| *** slaweq has joined #opendev | 13:43 | |
| mordred | corvus: also - I think one of the followup thoughts I had was maybe valid - which is that "base" might be still a bit off - and we might want to split the roles up in to "seriously this just wants to apply everywhere buthas nothing to do with services" like user accounts - and keep that stuff as base and remove the dependencies on it - and then there's probably only one or two roles in base that have | 13:45 |
| mordred | potential service-impact, like iptables - and we could focus on getting those into something so that they run at the right time | 13:45 |
| corvus | mordred: oh yeah that's a good point | 13:48 |
| corvus | i think that could help a lot | 13:48 |
| mordred | corvus: in fact - looking through base, I think literally iptables is the only thing where it's important that we run it before a service | 13:49 |
| mordred | everything else is just general operational stuff | 13:49 |
| corvus | yeah, i'm feeling good about this | 13:50 |
| mordred | corvus: so - I think that means we might just want to add iptables to each service playbook | 13:51 |
| mordred | no need to have a separate playbook for that | 13:51 |
| corvus | ++ | 13:52 |
| *** slaweq has quit IRC | 13:56 | |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 14:12 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Fix py27 tox https://review.opendev.org/731187 | 14:13 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Clean up base playbook https://review.opendev.org/730985 | 14:22 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Move base roles into a base subdir https://review.opendev.org/730937 | 14:22 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Run iptables in service playbooks instead of base https://review.opendev.org/730999 | 14:22 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Split inventory into multiple dirs and move hostvars https://review.opendev.org/730991 | 14:22 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Move rackspace repo management to the repos role https://review.opendev.org/731224 | 14:22 |
| mordred | corvus: ok. I think that might actually not be terrible ^^ | 14:22 |
| mordred | corvus: ALSO - I realized that since we always run install-ansible - we could probably move all of the roles from playbooks/roles to just roles/ | 14:22 |
| mordred | corvus: or we could just add /home/zuul/src/opendev.org/opendev/system-config/roles to the roles path in ansible.cfg - but I think structurally things a little bit different than they were when we started this | 14:24 |
| corvus | mordred: we put them in playbook/roles because we didn't want them to be "public" roles | 14:25 |
| corvus | (ie, zuul roles) | 14:25 |
| mordred | corvus: oh! I just had another realization (slightly based on that) ... | 14:27 |
| mordred | corvus: we could just run the install-ansible role in playbooks/zuul/run-production-playbook.yaml and ditch the additional install-ansible job | 14:28 |
| openstackgerrit | Merged openstack/project-config master: Match Debian in AFS slug generation https://review.opendev.org/731009 | 14:29 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 14:35 |
| mordred | AJaeger: zomg. are you tacking that beast? ^^ awesome! | 14:36 |
| AJaeger | mordred: see my message on #zuul | 14:36 |
| AJaeger | mordred: if you would have described it as beast earlier, I might not have started ;) | 14:37 |
| AJaeger | mordred: and no promise to finish ;) But moving it forward a bit | 14:38 |
| mordred | AJaeger: awesome. not a beast - a very cuddly bunny!!! | 14:38 |
| *** hashar has joined #opendev | 14:38 | |
| AJaeger | too late ;) | 14:38 |
| openstackgerrit | Merged openstack/project-config master: build and publish wheels for more distributions (x86-64) https://review.opendev.org/730323 | 14:46 |
| openstackgerrit | Merged openstack/project-config master: build and publish wheels for aarch64 architecture https://review.opendev.org/730342 | 14:46 |
| hrw | ;) | 14:47 |
| hrw | morning | 14:47 |
| *** slaweq has joined #opendev | 14:47 | |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Fix py27 tox https://review.opendev.org/731187 | 14:49 |
| jrosser | does python2 get installed into the ubuntu focal CI images? | 14:52 |
| openstackgerrit | Matt McEuen proposed openstack/project-config master: Project creation request: airship/charts https://review.opendev.org/731236 | 14:55 |
| fungi | jrosser: as far as i know we don't install python2.7 packages by default in our bionic or focal images, but i'll check an image build log to confirm... why do you ask? | 14:55 |
| jrosser | i have a suspicion that the answer is yes, but theres a chance it could be my mistake putting it there | 14:55 |
| jrosser | i see a bunch of python2 things here https://storage.bhs.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_c0a/723597/9/check/openstack-ansible-deploy-aio_metal-ubuntu-focal/c0a2e23/logs/ubuntu-apt-list-installed-host-08-15-05.txt | 14:56 |
| fungi | jrosser: actually it seems to get pulled in as a dependency of something else we're installing in the images... see 2020-05-25 20:33:05.553 in https://nb01.opendev.org/ubuntu-focal-0000009828.log | 14:57 |
| openstackgerrit | Merged zuul/zuul-jobs master: Remove unused tox_envlist in fetch-subunit-output https://review.opendev.org/729348 | 14:58 |
| fungi | jrosser: it must be a direct or indirect dependency of one of these: git util-linux iputils-ping python3-dev curl lvm2 gnupg2 strace dnsutils cron wget isc-dhcp-client e2fsprogs ifupdown net-tools unbound adduser acpid build-essential uuid-runtime apt-transport-https haveged linux-image-generic ca-certificates openssh-client iptables-persistent python3-venv python-dev tcpdump parted openssh-server locales | 14:58 |
| fungi | iproute2 rsyslog cloud-utils initramfs-tools at traceroute coreutils rsync | 14:58 |
| fungi | my bet is python-dev is at fault | 14:58 |
| jrosser | fungi: ah interesting, for focal there are a bunch of interesting/unfortunate side effects | 14:58 |
| mordred | fungi: yeah - python-dev would do it | 14:59 |
| jrosser | if ansible decides that py2 is the interpreter of choice, all sorts of bad times happen with missing python-apt, and so on | 14:59 |
| mordred | I think we can blank it out in our infra-package-needs - we're doing that for centos8 | 15:00 |
| fungi | https://opendev.org/openstack/project-config/src/branch/master/nodepool/elements/infra-package-needs/package-installs.yaml#L8 | 15:01 |
| openstackgerrit | Monty Taylor proposed openstack/project-config master: Don't install python-dev on focal https://review.opendev.org/731237 | 15:01 |
| mordred | fungi, jrosser ^^ maybe like that | 15:01 |
| fungi | https://opendev.org/openstack/project-config/src/branch/master/nodepool/elements/infra-package-needs/pkg-map#L11-L14 | 15:01 |
| fungi | adding there | 15:02 |
| mordred | ++ | 15:02 |
| jrosser | out of interest would you expect it to be present on bionic as well? | 15:02 |
| fungi | jrosser: yep, for the same reason | 15:03 |
| fungi | jrosser: though dropping it from bionic will likely be a harder proposition | 15:03 |
| jrosser | indeed | 15:03 |
| fungi | people have been relying on those images for over two years, so there are likely lots of jobs assuming python2.7 will be present on them | 15:03 |
| mordred | yeah - and it will have been there for longer, so people will likely have managed it already | 15:03 |
| mordred | yeah | 15:03 |
| fungi | also, 731237 may not be sufficient on its own, if any of the other packages in there also depend on python2.7 on focal | 15:04 |
| jrosser | for bionic there are suitable versions of things like python-apt and python2-lxc from the bionic repo, so it can be worked around | 15:04 |
| jrosser | but thats not the case for focal | 15:04 |
| jrosser | python-apt is particularly annoying as there is only a very old version installable with pip | 15:05 |
| *** priteau has joined #opendev | 15:07 | |
| clarkb | why is ansible using python2 instead of 3 if both are present? | 15:08 |
| clarkb | that also seems like an ansible bug | 15:08 |
| *** melwitt has joined #opendev | 15:09 | |
| clarkb | (we should efinitely fix focal, but maybe we should file a bug with ansible to prefer 3 if present?) | 15:09 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Fix py27 tox https://review.opendev.org/731187 | 15:09 |
| mordred | clarkb: it's a behavior change that will happen in a forthcoming release | 15:10 |
| mordred | clarkb: ansible currently will emit a warning about it | 15:10 |
| clarkb | mordred: thats completely unexpected "auto" behavior | 15:10 |
| hrw | thanks for merging wheel jobs | 15:10 |
| clarkb | I Guess that is even more reason to not rely on the version of python ansible is using if you care about python versions | 15:11 |
| mordred | clarkb: ah - it might just be on 16.04 | 15:11 |
| frickler | fungi: mordred: I verified that none of the other pkgs will pull in py2 on focal | 15:11 |
| mordred | clarkb: [DEPRECATION WARNING]: Distribution Ubuntu 16.04 on host afs01.ord.openstack.org should use /usr/bin/python3, but is using /usr/bin/python for | 15:11 |
| mordred | backward compatibility with prior Ansible releases. A future Ansible release will default to using the discovered platform python for this | 15:11 |
| mordred | host. See https://docs.ansible.com/ansible/2.9/reference_appendices/interpreter_discovery.html for more information. This feature will be | 15:11 |
| mordred | removed in version 2.12. Deprecation warnings can be disabled by setting deprecation_warnings=False in ansible.cfg. | 15:11 |
| fungi | frickler: awesome, i went through the build log checking the same, python-xml is no longer present on focal so looks like we don't install it | 15:12 |
| mordred | clarkb: that's the warning we get about this currently | 15:12 |
| clarkb | mordred: oh wow the reason is "auto" isn't actually the default. auto_legacy is | 15:12 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Fix py27 tox https://review.opendev.org/731187 | 15:12 |
| clarkb | so if we had nodepool force "auto" over "auto_legacy" we'd get the correct behavior? | 15:12 |
| mordred | clarkb: yeah - I think they did that to allow people some time to transition | 15:12 |
| mordred | clarkb: I think so, yes | 15:12 |
| mordred | clarkb: and I think that's a good idea - but I thought we _did_ set auto already | 15:13 |
| *** Open10K8S has quit IRC | 15:16 | |
| *** Open10K8S has joined #opendev | 15:16 | |
| mordred | fungi, frickler: the project-config-dib job seems unhappy - I seem to remember a patch from ianw recently to dib about adding yamllint - perhaps that for some reason isn't gettign yamllint installed for us? | 15:21 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 15:22 |
| openstackgerrit | Monty Taylor proposed openstack/project-config master: Don't install python-dev on focal https://review.opendev.org/731237 | 15:25 |
| mordred | fungi, frickler, clarkb: ^^ that should fix the job for project-config as well now | 15:25 |
| mordred | ianw: when you're up - yamllint is now needed by dib-lint, but is only in test-requirements.txt of dib, so external use of yamllint by people will now break | 15:26 |
| *** ykarel is now known as ykarel|away | 15:27 | |
| clarkb | mordred: beacuse we are running dib lint and that deps on yamllint? | 15:27 |
| openstackgerrit | Monty Taylor proposed openstack/diskimage-builder master: Add dependency on yamllint https://review.opendev.org/731245 | 15:28 |
| mordred | clarkb: yup. that ^^ should fix it dib-side | 15:28 |
| mordred | I'm starting to feel like dib-lint should be split into its own package | 15:28 |
| openstackgerrit | Merged zuul/zuul-jobs master: Fix py27 tox https://review.opendev.org/731187 | 15:28 |
| mordred | as this is now 2 linters that you have to install to build images | 15:29 |
| clarkb | mordred: ya but what is the cost of that | 15:29 |
| clarkb | its there ina virtualenv or a docker image and its not doing anything if you don't use it | 15:29 |
| clarkb | if we're worried about disk space we should stop using docker | 15:30 |
| mordred | I believe we had an issue a little whiel ago where dib was pulling in flake8 and that was conflicting somewhere | 15:30 |
| mordred | I'm not worried about disk space | 15:30 |
| mordred | I'm worried about suitability of dependencies and conflicts at times where you don't expect it | 15:30 |
| jrosser | fungi: for the focal python thing, python-is-python3 may be needed | 15:30 |
| clarkb | mordred: I think that conflict was a bug itself, not necessarily beacuse we have deps | 15:31 |
| clarkb | and that specific issue was addressed? | 15:31 |
| clarkb | jrosser: you're saying that removing python2 and `python` isn't sufficient for some reason? | 15:31 |
| mordred | clarkb: https://review.opendev.org/#/c/712778/ it was this | 15:32 |
| jrosser | https://packages.ubuntu.com/focal/python-is-python3 | 15:32 |
| clarkb | mordred: ya we were installing hacking without using hacking | 15:32 |
| clarkb | jrosser: right I'm asking why `python3` isn't sufficient I guess | 15:32 |
| mordred | clarkb: yeah. well - that was because nodepool depends on dib and dib depended on hacking which caused hacking to be installed in nodepool unexpectedly. flake8 and yamllint aren't as invasive, but still, as a person installing a tool to build diskimages, I don't know that I expect it to install linting tools. it's obviously not a hill I'm going to die on - but I think it's inappropriate | 15:33 |
| clarkb | jrosser: mordred's change https://review.opendev.org/#/c/731237/2/nodepool/elements/infra-package-needs/pkg-map should remove python2 and `python` | 15:33 |
| jrosser | ok | 15:34 |
| fungi | jrosser: i strongly recommend against the python-is-python3 package, it violates guidance from the python community on how you should identify your python and python3 interpreters | 15:34 |
| fungi | (basically there are two programming languages, one called "python" and one called "python3" and their interpreters are named accordingly) | 15:35 |
| clarkb | also I think we need to push more for "be specific about the python you want if it matters to you" | 15:35 |
| clarkb | we've run into this with tox | 15:35 |
| clarkb | we're now running into this with ansible apparently | 15:35 |
| clarkb | its unlikely we'll be able to make everyone happy all the time purely through defaults | 15:35 |
| fungi | yes, ideally people would use `python2` to disambiguate, but historically `python` means python<3 (unless you're using arch) | 15:36 |
| clarkb | fungi: well in this case I mean use `python3` if you need python3 too :) | 15:36 |
| clarkb | (it applies in both directions) | 15:36 |
| fungi | absolutely | 15:37 |
| fungi | you should definitely call `python3` if you don't want to run 2.7 | 15:38 |
| AJaeger | fungi, want toe merge https://review.opendev.org/#/c/731072/ to update grafana for the new wheels, please? | 15:38 |
| *** priteau has quit IRC | 15:42 | |
| clarkb | I'm trying to debug meetpad cacti before booting a new jvb, just so that we have additioanl data as we test and bring that up | 15:44 |
| clarkb | the poller is timing out takling to meetpad's snmpd | 15:44 |
| clarkb | any idea where we configure snmp connection details for cacti? | 15:44 |
| clarkb | (I'd like to snmpwalk directly and debug from there) | 15:44 |
| openstackgerrit | Merged zuul/zuul-jobs master: fetch-tox-output: empty envlist should behave like tox -e ALL https://review.opendev.org/730334 | 15:46 |
| fungi | first, you're sure the firewall is allowing snmp/udp (or tcp if we're using v3 over tcp)? | 15:47 |
| fungi | i think we just rely on firewall rules rather than setting secret community strings | 15:47 |
| clarkb | fungi: yes I checked the firewall yseterday and udp port 161 is allowed from cacti server | 15:47 |
| clarkb | fungi: ya I thought so too but snmpwalk insists on a value for that | 15:48 |
| clarkb | maybe I give it an empty string | 15:48 |
| fungi | so probably the default of "public" | 15:48 |
| fungi | also i think you need to use v2a if you don't want to have to set username and password | 15:48 |
| corvus | v2c ? | 15:48 |
| fungi | er, yep, v2c sorry | 15:48 |
| fungi | anyway, point being v3 no longer relies on community strings as an acl | 15:49 |
| fungi | so if you're trying to use v3 that could explain why snmpwalk wants un/pw | 15:49 |
| clarkb | ya I've forced v2c then it wanted a community string | 15:50 |
| clarkb | give it an empty community string seems to timeout. Using 'public' doesn't | 15:50 |
| corvus | public is correct | 15:51 |
| openstackgerrit | Merged zuul/zuul-jobs master: ensure-haskell-stack: Make sure dnf copr is installed https://review.opendev.org/731173 | 15:51 |
| corvus | clarkb: http://cacti.openstack.org/cacti/host.php?action=edit&id=629 | 15:51 |
| openstackgerrit | Merged openstack/project-config master: AFS grafana : add wheel volumes https://review.opendev.org/731072 | 15:51 |
| *** factor has joined #opendev | 15:51 | |
| *** _mlavalle_1 has quit IRC | 15:53 | |
| clarkb | corvus: I need to login there and am having a hard time finding where we've stashed those credentials | 15:55 |
| corvus | clarkb: usual place | 15:55 |
| corvus | cacti admin | 15:55 |
| clarkb | gotcha so the configs are all in the databse I guess | 15:59 |
| clarkb | and they look correct | 16:00 |
| corvus | yeah, i don't know why the php code is having trouble with it when snmpwalk isn't | 16:00 |
| clarkb | on the meetpad side I see snmpd recording connections from ipv6 cacti02 address | 16:01 |
| clarkb | but its not saying much more there | 16:01 |
| clarkb | if I do an snmpwalk over ipv4 the meetpad side logging looks similar | 16:02 |
| clarkb | what if packets from meetpad to cacti over ipv6 are bitbucketing | 16:02 |
| * clarkb does some connection testing | 16:03 | |
| clarkb | I think that may be it | 16:03 |
| clarkb | wget http://cacti.openstack.org from meetpad is failing to connect | 16:03 |
| clarkb | and since snmp is all udp we don't get the connection errors, we just see packets poof and then timeout | 16:04 |
| *** mlavalle has joined #opendev | 16:04 | |
| clarkb | alright knowing that maybe I should jsut boot a jvb01 and not worry about it for now | 16:06 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Move base roles into a base subdir https://review.opendev.org/730937 | 16:06 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Move rackspace repo management to the repos role https://review.opendev.org/731224 | 16:06 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Run iptables in service playbooks instead of base https://review.opendev.org/730999 | 16:06 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Split inventory into multiple dirs and move hostvars https://review.opendev.org/730991 | 16:06 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Override bridge hostvars directly https://review.opendev.org/731258 | 16:06 |
| clarkb | wget eventually falls back to ipv4 and it works | 16:06 |
| mordred | corvus: 731258 should show us whether the "just override hostvars" thing will work | 16:06 |
| fungi | clarkb: yeah, sounds like one of the (many) ipv6 connectivity issues we see between specific instances in rackspace | 16:06 |
| corvus | mordred: cool, i'm working through the stack now | 16:06 |
| clarkb | I could put meetpad's current ipv4 address into cacti via the admin console if we want | 16:07 |
| mordred | corvus: cool - I think I just fixed all of the test errors | 16:07 |
| fungi | clarkb: cacti seems to be impacted more than other servers, but thinking about it that's probably just because it's one of the only systems we have which needs to connect to most other systems | 16:07 |
| clarkb | we'd have to remember to update that should the server move | 16:07 |
| clarkb | fungi: that and it doesn't seem to failover to ipv4 like just about everything else does | 16:07 |
| corvus | oh, maybe the php stuff doesn't failover but the cli does | 16:08 |
| clarkb | corvus: ya thats what I'm starting to think is happening | 16:08 |
| fungi | interesting | 16:08 |
| corvus | huh, though according to http://cacti.openstack.org/cacti/settings.php it looks a lot like we should be using net-snmp and not embedded php | 16:09 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 16:10 |
| *** LuckyClover has joined #opendev | 16:13 | |
| clarkb | jvb01 is now launching | 16:14 |
| clarkb | we can also try a reboot of meetpad and/or cacti to see if that unsticks ipv6 routing? | 16:15 |
| clarkb | cacti -> meetpad is working, we see those packets arrive on meetpad. But meetpad -> cacti doesn't appear to | 16:15 |
| *** rpittau is now known as rpittau|afk | 16:19 | |
| openstackgerrit | Merged zuul/zuul-jobs master: tox: envlist bugfixes https://review.opendev.org/730381 | 16:20 |
| corvus | mordred: comment on 730999 | 16:22 |
| *** dtantsur is now known as dtantsur|afk | 16:26 | |
| *** hashar has quit IRC | 16:28 | |
| corvus | mordred: and i think i don't understand the last 2 changes: what makes the fake-private hostvars gain precedence over the public hostvars? aren't the public hostvars still in a playbook-adjacent inventory directory? | 16:31 |
| openstackgerrit | Clark Boylan proposed opendev/system-config master: Add a jvb01 server https://review.opendev.org/731268 | 16:32 |
| mordred | corvus: no, they aren't - they're moved to the inventory directory | 16:33 |
| mordred | corvus: I'm not sure the fake-private _will_ have precedence - but I'm pretty sure inventories are processed in order, so making them have precedence should just be a matter of reordering them int he config file if they don't | 16:33 |
| corvus | mordred: ooh i see sorry | 16:33 |
| openstackgerrit | Clark Boylan proposed opendev/zone-opendev.org master: Add jvb01 to dns https://review.opendev.org/731269 | 16:34 |
| corvus | mordred: should inventory/service be added to ansible.cfg? | 16:34 |
| clarkb | infra-root I think ^ those two jvb changes can land in any order as we don't need to set up LE certs or anything for jvb aiui | 16:35 |
| clarkb | the server is up though and ready for ansibling I think | 16:35 |
| fungi | clarkb: the other times we've seen one of those v6 connectivity failures go away have been after host migrations | 16:35 |
| mordred | corvus: we copy them into /etc/ansible/hosts/{base,service} in the install-ansible role | 16:35 |
| fungi | clarkb: so a reboot probably won't do much (but can't hurt) | 16:35 |
| mordred | corvus: BUT - we could also not do that and just change the ansible.cfg to point to ~zuul/src/opendev.org/opendev/system-config/inventory/{base,service} - I'm torn on which is better | 16:36 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 16:36 |
| *** hashar has joined #opendev | 16:36 | |
| corvus | mordred: /etc/ansible/hosts is our private git repo; it does feel a bit weird to do that.... | 16:36 |
| corvus | mordred: i know we have some non-git-tracked files in there now, but this would be a lot more... | 16:37 |
| clarkb | oh infra-root I've got to add the secret data to hiera for jvb01. | 16:38 |
| clarkb | doing that now | 16:38 |
| corvus | mordred: i think it's something we could get used to, but it may be easier to think about if it were in the config file? i am also torn | 16:38 |
| corvus | clarkb: what do we need that's new? | 16:39 |
| mordred | corvus: yeah - same. should I rework it to not copy and instead just point the config? | 16:40 |
| clarkb | corvus: we need to put the meetpad_jvb_password in the jvb group vars | 16:40 |
| clarkb | corvus: I've done that if you want to git log -1 -p in ansible hosts on bridge | 16:40 |
| mordred | corvus: pointing config might be more similar to the current behavior of whatever the vars are in the system-config repo being the ones that apply | 16:41 |
| corvus | mordred: 1 sec | 16:47 |
| *** ysandeep is now known as ysandeep|away | 16:49 | |
| corvus | mordred: so our current sitaution is public > private because public is playbook adjacent. 730991 makes public/base > public/service > private (because of the order in ansible.cfg: /etc/ansible/hosts/base, /etc/ansible/hosts/service, /etc/ansible/hosts). the alternative would be to do the same thing but without copying the files (so system-config/.../base, system-config/.../service, /etc/ansible/hosts) | 16:52 |
| clarkb | I've got to manage kids schooling for the next little bit but those two changes abvoe should be safe to land when reviewed | 16:52 |
| corvus | mordred: is that correct? | 16:52 |
| mordred | corvus: yeah- I thnik that is correct | 16:53 |
| mordred | corvus: we could also keep the public > private via ordering - and have our system-config jobs add an additional set of vars that are public overrides | 16:54 |
| mordred | if we don't want to invert the precedence | 16:54 |
| mordred | (but I think private > public is still likely fine yeah?) | 16:54 |
| corvus | mordred: yeah, i was just starting to think about that too, that's not a bad idea | 16:54 |
| mordred | corvus: nod | 16:54 |
| corvus | mordred: public > private does have the advantage of we can't accidentally override public stuff by sticking something in hiera | 16:55 |
| corvus | er, in private hostvars :) | 16:55 |
| corvus | mordred: i'm not sure i'm too worried about it tho | 16:55 |
| mordred | me either | 16:57 |
| mordred | I think copy vs not-copying is the bigger question | 16:57 |
| corvus | mordred: i think i like not-copying. it's 55%/45% for me :) | 17:00 |
| *** LuckyClover has quit IRC | 17:04 | |
| corvus | config-core: i'd like to draw your attention to this zuul change: https://review.opendev.org/718531 | 17:12 |
| corvus | if we merge that as written, people may have the option of per-branch change queues; if we want to disallow that, we'll need to define the queues in a config project first. | 17:13 |
| corvus | an earlier patchset made it so that could only be set in a config-project, but it wasn't clear that was the most consistent choice, so the current ps makes it settable by individual projects but overridable by a config-project. | 17:14 |
| clarkb | corvus: if we set nothing we retain the existing behavior? | 17:15 |
| clarkb | I think I'm ok with that if so | 17:15 |
| corvus | i see that as a small-to-moderate risk; i think we can set the flag to falso on the openstack integrated queue and probably generally be okay. | 17:15 |
| clarkb | then we can educate projects in what choice may be appropriate for them if they start setting the toggle | 17:15 |
| corvus | clarkb: correct -- and i don't think we expect to change that default. | 17:15 |
| corvus | note also that i expect this to apply to the upcoming bi-directional dependency work too | 17:16 |
| clarkb | this being new config object we can set in trusted config to enforce behavior? | 17:17 |
| clarkb | I expect that one we may wantto enforce | 17:17 |
| corvus | clarkb: yes, same mechanism / defaults / etc | 17:17 |
| corvus | so there won't be a way to force it globally, just on an individual queue level. but we can force the settings on the queues we care about. | 17:18 |
| corvus | if that's a non-starter for opendev, now would be a good time to find out :) | 17:19 |
| clarkb | I dont think its a non starter, but I do think projects inside and out of opemdev should have a hard think about changing the defaults | 17:19 |
| corvus | the documentation does mention the reasoning and potential pitfalls | 17:20 |
| clarkb | the branch thing is important if you intend to do any upgrade testing between branches. The mutual dependency issue is important to avoid for continuous deployment and bisectability. Neitherof those will be important to every git repo so having zuul make them configurable is good. But we want to do our best to avoid making those attractive toggles for people that just want to merge code and undermine zuul's | 17:21 |
| clarkb | purpose | 17:21 |
| clarkb | from opendev's perspective I think that means education more than anything else | 17:22 |
| corvus | yeah. i guess if we back ourselves into a corner and find that we need to set a tenant-level override, we can still do that in the future, and fairly easily. i think the harder thing to do would be to make the queue object into a config-project only object (like pipeline) as in tobiash's last patch. once it's out there at the untrusted level, it'll be hard to move. but if you assume all the actors are | 17:24 |
| corvus | reasonable, i think having it in untrusted feels like the right choice. | 17:24 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/project-config master: Try using vexxhost CA region https://review.opendev.org/731283 | 17:25 |
| mnaser | config-core: ^ if possible, we'd like to move out from sjc1 to mtl1 (to newer hardware too for capacity reasons) | 17:26 |
| fungi | it does seem like something which can be decided by each project. but with my tact sig hat on i would press openstack to stick with our prior conventions (no per-branch queues, no bi-directional dependencies) | 17:26 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 17:26 |
| fungi | mnaser: thanks for the heads up, taking a look now | 17:26 |
| openstackgerrit | Merged openstack/project-config master: Remove check/gate jobs for x/packstack https://review.opendev.org/730813 | 17:27 |
| clarkb | fungi: ya I think openstack shouldnt change frkm the defaults. But other opendev users might | 17:27 |
| jrosser | when would i see a focal CI image incorporating the python2 change? | 17:27 |
| fungi | mnaser: is ymq1 the same as mtl1? | 17:27 |
| mnaser | fungi: yeah | 17:27 |
| fungi | thanks | 17:27 |
| clarkb | C/C++ api changes can be rough with out some sort of override or mutual deps | 17:28 |
| openstackgerrit | Merged openstack/project-config master: Remove gate/check jobs for cloudpulse and cloudpulseclient https://review.opendev.org/731089 | 17:30 |
| corvus | clarkb: some people may go to space today | 17:31 |
| corvus | apparently they are driving to their rocket now | 17:31 |
| clarkb | corvus: yup. weatherus looking spotty but uf they launch it will be in about 3 hours aiui | 17:32 |
| fungi | mnaser: out of curiosity, where does the ymq abbreviation come from? (mtl is fairly obvious at least) | 17:33 |
| fungi | i'm guessing it's not an iata airport code, looks like the only ones in the montreal area are ymx and yul | 17:33 |
| corvus | fungi: https://en.wikipedia.org/wiki/List_of_airports_by_IATA_airport_code:_Y#cite_note-IATA-1 | 17:34 |
| fungi | mnaser: nevermind, looks like that's a regional iata code covering all of montreal | 17:34 |
| mnaser | fungi: yep :) | 17:35 |
| fungi | yep, that seems to confirm, thanks! | 17:35 |
| mnaser | YUL is "the" airport though, YMX was supposed to be the "big awesome future airport" that has kinda just become a cargo hub | 17:35 |
| fungi | yeah, i remember you showing me aerial photos of that one while we were at ocw | 17:37 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 17:42 |
| openstackgerrit | Colleen Murphy proposed openstack/diskimage-builder master: Pre-install xz package in opensuse chroot https://review.opendev.org/730992 | 17:45 |
| *** hashar has quit IRC | 17:53 | |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 17:53 |
| clarkb | mordred: corvus have a quick moment for the jvb server add in https://review.opendev.org/#/q/topic:jvb01+status:open | 17:57 |
| *** dpawlik has quit IRC | 17:59 | |
| openstackgerrit | Merged openstack/project-config master: Try using vexxhost CA region https://review.opendev.org/731283 | 18:03 |
| clarkb | jrosser: it takes about an hour to build the image then $time to upload it | 18:05 |
| clarkb | jrosser: that happens automatically every 24 hours but we can trigger it sooner if the change has landed | 18:05 |
| clarkb | (I odn't think it has merged yet) | 18:06 |
| jrosser | clarkb: ok, thankyou | 18:06 |
| openstackgerrit | Merged opendev/zone-opendev.org master: Add jvb01 to dns https://review.opendev.org/731269 | 18:07 |
| *** roman_g has quit IRC | 18:19 | |
| *** roman_g has joined #opendev | 18:20 | |
| *** roman_g has quit IRC | 18:20 | |
| openstackgerrit | Merged opendev/system-config master: Add a jvb01 server https://review.opendev.org/731268 | 18:21 |
| *** roman_g has joined #opendev | 18:21 | |
| clarkb | as a heads up I've had to accept the ssh host key for jvb01 as root on bridge | 18:26 |
| clarkb | I thought launch was going to do that for me, but I guess it doesn't? | 18:26 |
| clarkb | the job for meetpad hasn't run yet so hopefully when it gets there it will be good to go | 18:27 |
| corvus | clarkb: yeah, i think we expect it to accept, so it's worth looking into | 18:38 |
| openstackgerrit | Oleksandr Kozachenko proposed zuul/zuul-jobs master: Add the namespace in wait-for-pods role https://review.opendev.org/731292 | 18:38 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: WIP: Add CentOS-8 support https://review.opendev.org/731293 | 18:38 |
| openstackgerrit | Sorin Sbarnea (zbr) proposed zuul/zuul-jobs master: Bump ansible-lint to 4.3.0 https://review.opendev.org/702679 | 18:41 |
| clarkb | base failed | 18:44 |
| clarkb | it failed because of host key verification to jvb01 | 18:45 |
| clarkb | so that must've run before I managed to accept the keys | 18:45 |
| clarkb | mordred: corvus what is the correct way for me to rerun those jobs? | 18:45 |
| clarkb | is it zuul enqueue (or enqueue-ref)? | 18:46 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: WIP: Add CentOS-8 support https://review.opendev.org/731293 | 18:51 |
| openstackgerrit | Oleksandr Kozachenko proposed zuul/zuul-jobs master: Add the namespace in wait-for-pods role https://review.opendev.org/731292 | 18:51 |
| clarkb | infra-root does this look correct `zuul enqueue --pipeline deploy --change 731268,1 --trigger gerrit --tenant openstack --project opendev/system-config` | 18:54 |
| fungi | clarkb: yep | 18:56 |
| clarkb | thanks I'll run that now since I've addressed the host key thing | 18:56 |
| zbr | clarkb: https://review.opendev.org/#/c/729974/ please. | 18:58 |
| clarkb | zbr: why does that stop using raw strings? | 18:58 |
| clarkb | oh wait its the othe rway around | 18:59 |
| clarkb | I'm not sure thats correct | 18:59 |
| zbr | one thing for sure '\ ' is invalid in newer python | 18:59 |
| clarkb | zbr: it works for me, however I expected them to differ in behavior but they don't | 19:00 |
| clarkb | print('\ ') is valid python3 | 19:00 |
| zbr | is an invalid escape sequence | 19:00 |
| clarkb | it works fine if you do ^ on python3.8.2 | 19:00 |
| zbr | try with py38 or py39 and see what happens | 19:00 |
| clarkb | but it seems to be equivalent to r'\ ' | 19:01 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add updates mirror https://review.opendev.org/731305 | 19:01 |
| clarkb | probably because its not a proper escape sequence the \ becomes a literal | 19:01 |
| clarkb | and r'\ ' forces it to be literal | 19:02 |
| zbr | py38 throws SyntaxWarning on it | 19:02 |
| fungi | env PYTHONWARNINGS=default::DeprecationWarning python3 -c 'print("\ ")' | 19:02 |
| fungi | it's hidden if deprecation warnings aren't on | 19:02 |
| zbr | why are we having this conversation? there is at least one way to write it without any warnings. | 19:03 |
| clarkb | fungi: aha thanks | 19:03 |
| clarkb | zbr: becaus it literally works without issue for me | 19:03 |
| clarkb | zbr: and its a potentially functional not linting change | 19:04 |
| fungi | zbr: i was trying to explain how to reproduce the warning | 19:04 |
| clarkb | and I want to undersatnd what it is made | 19:04 |
| zbr | i usually setup warns: errors,... where I can. | 19:04 |
| zbr | ahh. sure. | 19:04 |
| clarkb | we have conversations about changes we make to ensure they do what we expect them to | 19:04 |
| zbr | basically python becomes more strict, which is good. | 19:04 |
| clarkb | this is an important piece of code ote nsure that gerrit groups get created properly. I want to make sure I understand it | 19:04 |
| zbr | yep, that is a sensitive line :D | 19:05 |
| clarkb | my understanding is that python3.8 is fine with it as is, but there is a deprecation warning if yo uturn them on so a future python version may make that an error. The code should be equivalent because '\ ' with the warning is treated as r'\ ' not ' ' | 19:06 |
| fungi | and yes, currently '\ ' and r'\ ' are treated the same, but eventually it's assumed their behaviors may need to diverge (hopefully after some period of not working at all) | 19:06 |
| clarkb | and all my testing between python2.7 and 3.8 shows they are equivalent | 19:06 |
| clarkb | we should update it to ensure we get what we want in python3.new | 19:06 |
| fungi | agreed | 19:07 |
| clarkb | and its understanding ^ that was missed for me as python3.8 and 2.7 say nothing about it | 19:07 |
| fungi | it got marked as deprecated syntax in 3.6 | 19:07 |
| fungi | but python doesn't show deprecation warnings by default | 19:08 |
| zbr | https://review.opendev.org/#/c/729328/ also has similar changes. | 19:08 |
| fungi | i add setenv = PYTHONWARNINGS=default::DeprecationWarning to the default [testenv] in tox.ini for my personal projects because i appreciate the verbosity and early deprecation warnings on those | 19:09 |
| fungi | otherwise you won't see them until they go from deprecated to broken | 19:09 |
| zbr | ideally we should do PYTHONWARNINGS=errors -- on those projects where we can afford it. | 19:10 |
| clarkb | it might be worth updating the test case we've got for new groups to ensure we are managing a group with a space in the name properly | 19:11 |
| clarkb | but I've approved the change as is | 19:11 |
| *** sshnaidm is now known as sshnaidm|afk | 19:12 | |
| zbr | the new flake8 added some very useful tests, already helped me uncover some real bugs. | 19:13 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: WIP: Add CentOS-8 support https://review.opendev.org/731293 | 19:21 |
| mordred | I like it when flake8 uncovers actual bugs | 19:23 |
| mordred | it's pleasing | 19:23 |
| openstackgerrit | Merged opendev/gerritlib master: Fixed linting issues https://review.opendev.org/729974 | 19:23 |
| clarkb | ansible is talking to jvb01 now in base | 19:27 |
| mordred | clarkb, fungi: corvus and I were having a discussion earlier about https://review.opendev.org/#/c/730991/ (which is the patch to move hostvars around) - before I do the next patchset, I'd like your input | 19:28 |
| mordred | we have 2 options - copy the hostvars from system-config into /etc/ansible/hosts/{subdir} - or configure ansible.cfg to point to ~zuul/src/opendev.org/opendev/system-config/inventory/{subdir} | 19:29 |
| mordred | corvus leans 55%/45% in favor of not copying - I think I'm similar | 19:30 |
| clarkb | currently we point at ~zuul right? | 19:30 |
| mordred | currently we use what's in ~zuul because of physical adjacency to the playbooks | 19:30 |
| clarkb | the upside to not copying is we keep the private data git manageable without copied in noise right? | 19:30 |
| clarkb | we can always git ignore those paths though I guess | 19:30 |
| mordred | clarkb: yes, that's right | 19:31 |
| mordred | we can - but having them there does potentially seem weird | 19:31 |
| clarkb | ya I think I prefer not copying in order to keep the git repo "clean" | 19:31 |
| corvus | yeah, i'd expect we'd git ignore; but there would be a cognitive overhead to knowing that | 19:31 |
| corvus | 'git grep' is going to trip me up there :) | 19:31 |
| mordred | yeah. :) | 19:31 |
| mordred | ok - I think that's enough leaning in a direction for me to work on the next iteration of the patch | 19:32 |
| *** hashar has joined #opendev | 19:41 | |
| corvus | clarkb: looks like docker isn't installed on jvb01; i guess something still went wrong there? | 19:45 |
| clarkb | corvus: only base has run so far | 19:46 |
| clarkb | corvus: waiting for the meetpad playbook to run whihc should pull in docker | 19:46 |
| clarkb | basically we got psat where we failed previously | 19:46 |
| corvus | ah k, so just slow :) | 19:46 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/project-config master: Encrease vexxhost CA region presence https://review.opendev.org/731314 | 19:53 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: WIP: Sync with configure-mirrors https://review.opendev.org/731293 | 19:59 |
| openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Add updates mirror https://review.opendev.org/731305 | 20:01 |
| clarkb | corvus: it should be installing docker any moment now | 20:07 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: WIP: Sync with configure-mirrors https://review.opendev.org/731293 | 20:08 |
| *** olaph has joined #opendev | 20:09 | |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Run iptables in service playbooks instead of base https://review.opendev.org/730999 | 20:11 |
| clarkb | jvb is up | 20:12 |
| clarkb | trying to see if it has connected to meetpad yet | 20:13 |
| clarkb | there is an established connection between them on port 5222 so I think the answer is yes | 20:13 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: WIP: Sync with configure-mirrors https://review.opendev.org/731293 | 20:13 |
| clarkb | anyone in a spot to do a test call? I'm in front of my tv looking at a rocket launch countdown | 20:13 |
| corvus | clarkb: yeah, just a minute | 20:15 |
| openstackgerrit | Merged openstack/project-config master: Encrease vexxhost CA region presence https://review.opendev.org/731314 | 20:15 |
| corvus | clarkb: meetpad.opendev.org/meetpad-test ? | 20:17 |
| corvus | they are not going to space today :( | 20:17 |
| clarkb | arg ya launch abort started | 20:17 |
| clarkb | corvus: ya I'll get on that meetpad in a couple minutes. Need to move to other machine with mic and get off couch | 20:17 |
| fungi | i should see if i can figure out meetpad on my workstation, since i managed to get it working from my netbook | 20:20 |
| openstackgerrit | Oleksandr Kozachenko proposed zuul/zuul-jobs master: Add namespace in the collect-k8s-logs role https://review.opendev.org/731319 | 20:21 |
| fungi | huh, i just noticed my powered usb3 hub has headphone and mic jacks too | 20:21 |
| clarkb | good news everyone | 20:35 |
| clarkb | it seems to work | 20:35 |
| clarkb | what that means is if we need to scale meetpad up during the ptg we can add a jvb02.opendev.org - jvb99.opendev.org (and potentially more if necessary) and once the meetpad service playbook runs on those newer hsots they should join up and start sharing load | 20:36 |
| clarkb | corvus and I were able to confirm via the jicofo logs on meetpad01 that we had multiple conference rooms and one was on meetpad01 jvb and the other on jvb01 jvb | 20:36 |
| clarkb | look for participant added strings for that | 20:36 |
| clarkb | the next thing to address there may be the cacti problems so that we can monitor load and demand better | 20:37 |
| clarkb | infra-root any thoughts on configuring cacti to use the ipv4 address for meetpad01 via the admin ui? | 20:37 |
| corvus | clarkb: that sounds like a not-terrible option all things considered | 20:39 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Split inventory into multiple dirs and move hostvars https://review.opendev.org/730991 | 20:43 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Override bridge hostvars directly https://review.opendev.org/731258 | 20:43 |
| mordred | clarkb: ++ | 20:43 |
| clarkb | ok I'll do that now | 20:43 |
| mordred | corvus: configuring it to point rather than copy I think turned out to not be very ugly | 20:43 |
| clarkb | currently its set to the host fqdn which we can set it back to at some point if ipv6 is happier | 20:43 |
| clarkb | we have cacti data for meetpad now | 21:10 |
| *** hashar has quit IRC | 21:14 | |
| clarkb | infra-root https://review.opendev.org/#/c/730862/ I'd be curious to know what others think about that change. It will switch us to using https for apt fetches and pip talking to our pypi mirrors. I think that is a good thing since verification outside of that is basically not a thing for us right now | 21:17 |
| mordred | clarkb: I thnik it's a good idea - we're pretty confident that it's going to work properly right? | 21:20 |
| clarkb | mordred: I think so. We tested the apt/yum/emerge side in zuul-jobs functional testing, then used base-test to ensure zuul-jobs unittests worked with https (so thats apt + pypi coverage) | 21:21 |
| clarkb | probably the biggest gap is with zypper and yum/dnf but they do https a lot more than apt has I think | 21:21 |
| clarkb | (and they were all tested in zuul-jobs functional jobs intsalling emacs so should work) | 21:21 |
| mordred | clarkb: yah - apt has had the signed repos and mirror farms for forever so ssl adoption has been much slower | 21:23 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Clean up base playbook https://review.opendev.org/730985 | 21:28 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Move base roles into a base subdir https://review.opendev.org/730937 | 21:28 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Move rackspace repo management to the repos role https://review.opendev.org/731224 | 21:28 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Run iptables in service playbooks instead of base https://review.opendev.org/730999 | 21:28 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Split inventory into multiple dirs and move hostvars https://review.opendev.org/730991 | 21:28 |
| openstackgerrit | Monty Taylor proposed opendev/system-config master: Override bridge hostvars directly https://review.opendev.org/731258 | 21:28 |
| *** slaweq has quit IRC | 21:50 | |
| *** slaweq has joined #opendev | 21:52 | |
| *** icarusfactor has joined #opendev | 22:08 | |
| openstackgerrit | Nate Johnston proposed openstack/project-config master: Retire the Congress project https://review.opendev.org/721742 | 22:09 |
| *** factor has quit IRC | 22:10 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: shake-build: add shake build system job https://review.opendev.org/731365 | 22:21 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: shake-build: add shake build system job https://review.opendev.org/731365 | 22:22 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: WIP: shake-build: add shake build system job https://review.opendev.org/731365 | 22:34 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: WIP: shake-build: add shake build system job https://review.opendev.org/731365 | 22:46 |
| *** tkajinam has joined #opendev | 22:52 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: shake-build: add shake build system job https://review.opendev.org/731365 | 23:04 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: shake-build: add shake build system job https://review.opendev.org/731365 | 23:14 |
| *** tosky has quit IRC | 23:30 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: shake-build: add shake build system job https://review.opendev.org/731365 | 23:40 |
| openstackgerrit | Matthew Thode proposed openstack/diskimage-builder master: use stage3 instead of stage4 for gentoo builds https://review.opendev.org/717177 | 23:50 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!