Friday, 2021-04-23

openstackgerritClark Boylan proposed opendev/system-config master: Test if gerrit v3.2.8 specifically is happy with selenium
zigoclarkb: fungi:
clarkbianw: ^ thats another thought. If that passes testing then it would strongly imply something has very recently changed in gerrit00:03
zigoWhere can we see the ansible facts?!?00:04
clarkbzigo: that seems to confirm my suspicion00:04
clarkbzigo: there ^00:04
clarkbit says ansible_distribution_release is n/a00:04
zigoAh yeah ...00:04
zigoSaw it.00:04
zigoIs there a way to override this?00:04
fungii wonder if newer ansible has it correct00:04
zigoIt probably will also be an issue when running the puppet thingy ...00:04
clarkbfungi: it seems to say it is reading from /etc/os-release00:05
zigoIf I'm not mistaking, there's an ansible part involved there too.00:05
clarkbansible_distribution_file_path: /etc/os-release I read that to say "I sourced this info from here"00:05
fungiwe might be able to create or overwrite /etc/os-release in the debian-minimal dib element?00:06
clarkbfungi: not a bad idea00:06
zigo$ cat /etc/os-release  | grep PRETTY_NAME00:06
zigoPRETTY_NAME="Debian GNU/Linux bullseye/sid"00:06
zigoThat's what I have (on my laptop...).00:06
zigoMr. Ansible is being nasty ... it should know better ! :)00:07
clarkbhrm maybe it is a parsing issue in ansible then?00:07
ianw104.130.70.62 is a bullseye node00:07
clarkbzigo: if you run ansible against your laptop to get facts you should be able to see what it does00:07
zigoI believe it should be confused by the "/sid" added at the end until bullseye is released.00:07
ianwyes, os-release on that node is the same as zigos00:08
zigoclarkb: How one may run ansible's facter?00:08
clarkbzigo: I think it is called the setup module in ansible so you do something like ansible -i localhost, -m setup ?00:08
clarkbVERSION_CODENAME=buster <- that is what buster says00:09
clarkbianw's example bullseye node does not have a VERSION_CODENAME netry00:09
zigoansible: error: the following arguments are required: pattern00:09
clarkbansible -i localhost, localhost -m setup ?00:09
clarkbanyway I think the reason it breaks is there is no VERSION_CODENAME00:10
clarkbso it gets recorded as n/a00:10
clarkbVERSION_ID="10" and VERSION="10 (buster)" are also in my buster file00:11
zigoOn my laptop it does work !00:11
clarkbzigo: does your laptop have VERSION_* entries in /etc/os-release?00:11
clarkbhuh neat00:11
ianw2.9.18 fails for me, installing a later one00:12
clarkbianw: re Gerrit I've just noticed that system-config-run-review-3.2 isn't running against my change to force v3.2.8 across the board. We just seem to build an image there. Interesting.00:13
clarkbI'll push a new ps to that that forces the job to run too00:13
zigoThat's what is on my laptop ...00:13
*** hamalq has quit IRC00:13
ianwok i still get n/a with 2.10.800:14
openstackgerritClark Boylan proposed opendev/system-config master: Test if gerrit v3.2.8 specifically is happy with selenium
clarkbnow we should run the run job as well as the build00:14
zigoAnsible on Bullseye is 2.9.16+dfsg-1.100:14
zigoNo Debian specific patches (I just checked).00:15
zigoWell, none relevant to that issue we're discussing ...00:15
ianwstill fails with devel for me00:16
* zigo got to go sleep00:18
zigo2:20 am here ...00:18
ianwzigo: no worries, i'll have a poke and see what i can find00:19
ianw# lsb_release00:31
ianwNo LSB modules are available.00:31
ianwthat is i think the root cause here00:31
fungiinterestingly, `lsb_release -a` probably works?00:52
fungion my systems, `lsb_release` gives the same "No LSB modules are available." message but `lsb_release -a` reports (among other things) "Codename: bullseye"00:52
ianwfungi: indeed -- Codename:n/a01:23
ianwprint('Codename:\t%s' % distinfo.get('CODENAME', 'n/a'))01:24
ianwi feel like there might be a recursive loop here :)01:24
*** brinzhang0 has joined #opendev01:28
*** brinzhang_ has quit IRC01:31
*** slittle1 has quit IRC01:33
ianwclarkb: well, passed everything ...01:37
ianwi'll run a recheck and see01:37
*** xinliang has joined #opendev01:48
fungiinterestingly, i have VERSION_CODENAME=bullseye in /etc/os-release but technically i'm on sid02:00
fungibase-files (11.1) unstable; urgency=medium02:02
fungiAdd VERSION_CODENAME to os-release.02:03
fungifrom /usr/share/doc/base-files/changelog.gz02:03
fungilooks like 11 is what's in bullseye right now, not 11.1 yet02:04
fungiMigration status for base-files (11 to 11.1): BLOCKED: Needs an approval (either due to a freeze, the source suite or a manual hint)02:05
fungiso i think it's likely to work once base-files 11.1 migrates from sid to bullseye02:06
ianwi think i see what's going on here, it is a recursive problem02:53
ianw# apt-cache policy02:55
ianw 500 bullseye/main amd64 Packages02:56
ianw     release o=Debian,n=bullseye,c=main,b=amd6402:56
ianw     origin mirror.dfw.rax.opendev.org02:56
ianwi don't know why, but this does not have a "label" field02:56
ianw v
ianwthis bug -> explains why lsb_release won't look at /etc/debian_version for lsb_release -c02:59
openstackDebian bug 845651 in lsb-release "lsb-release: --codename returns n/a on stretch without apt sources configured" [Important,Open]02:59
ianwit falls through to looking at "apt-cache policy"03:00
ianw*however*, it will not parse the line "release o=Debian,n=bullseye,c=main,b=amd64" because it doesn't have "l=Debian"03:00
openstackgerritIan Wienand proposed opendev/system-config master: reprepro : add labels
*** ykarel|away has joined #opendev04:18
openstackgerritXinliang Liu proposed opendev/system-config master: Enable openEuler mirroring
xinliangianw, please help to review openEuler mirroring patch thanks, have updated it :
*** vishalmanchanda has joined #opendev04:36
*** xinliang has quit IRC04:55
*** ykarel_ has joined #opendev04:55
*** ysandeep|away is now known as ysandeep04:58
*** ykarel|away has quit IRC04:58
openstackgerritIan Wienand proposed openstack/diskimage-builder master: Set bullseye
*** snapdeal has joined #opendev05:11
openstackgerritIan Wienand proposed openstack/diskimage-builder master: debian-minimal: Set bullseye version
ianwfungi / zigo : an examination of what i think is going on ... :/05:12
*** ykarel__ has joined #opendev05:23
*** ykarel_ has quit IRC05:23
*** priteau has joined #opendev05:29
*** ykarel_ has joined #opendev05:41
*** ykarel_ is now known as ykarel05:43
*** ykarel__ has quit IRC05:43
*** brinzhang_ has joined #opendev06:12
*** brinzhang0 has quit IRC06:15
*** slaweq has joined #opendev06:20
*** ralonsoh has joined #opendev06:26
*** sboyron has joined #opendev06:30
*** amoralej|off is now known as amoralej06:43
ianwi think the dib gate is hung up on something to do with xenial+glean+pbr but i don't have time to diagnose it now06:44
*** eolivare has joined #opendev07:00
*** ykarel has quit IRC07:07
*** ykarel has joined #opendev07:10
*** rpittau|afk is now known as rpittau07:12
*** ykarel_ has joined #opendev07:12
*** ykarel has quit IRC07:15
*** whoami-rajat_ has joined #opendev07:17
*** whoami-rajat_ is now known as whoami-rajat07:25
zigoianw: Thanks, that's a very interesting read.07:32
openstackgerritAlbin Vass proposed zuul/zuul-jobs master: Roles to create, cleanup and promote snapshots in ec2
*** snapdeal has quit IRC07:42
*** hrw has joined #opendev07:43
*** ykarel_ has quit IRC07:45
*** andrewbonney has joined #opendev07:49
*** tosky has joined #opendev07:53
*** jpena|off is now known as jpena07:54
*** ykarel has joined #opendev07:55
*** ykarel has quit IRC08:04
*** ykarel has joined #opendev08:06
*** hemanth_n has joined #opendev08:16
*** dtantsur|afk is now known as dtantsur08:17
fricklerinfra-root: I received feedback from the neutron team that they moved back from meetpad to zoom because of repeated audio problems I also saw some note in the nova channel earlier that chinese participants still seem to have issues08:56
fricklerso maybe we can try to do some more testing and see how to improve this for the next event08:58
*** ysandeep is now known as ysandeep|lunch09:11
*** dtantsur is now known as dtantsur|bbl09:30
*** ykarel_ has joined #opendev09:39
*** ykarel has quit IRC09:41
*** ykarel_ is now known as ykarel09:42
*** ysandeep|lunch is now known as ysandeep09:55
*** ykarel_ has joined #opendev10:15
*** ykarel has quit IRC10:18
*** ykarel__ has joined #opendev10:31
*** ykarel_ has quit IRC10:33
*** ykarel_ has joined #opendev10:36
*** ykarel_ is now known as ykarel10:37
*** ykarel__ has quit IRC10:39
*** jpena is now known as jpena|lunch11:30
fungiwe did have multiple chinese participants in the opendev session11:33
*** snapdeal has joined #opendev11:45
openstackgerritAnanya proposed opendev/elastic-recheck master: WIP: Run elastic-recheck container
*** hemanth_n has quit IRC12:25
*** dtantsur|bbl is now known as dtantsur12:29
*** amoralej is now known as amoralej|lunch12:33
*** jpena|lunch is now known as jpena12:33
openstackgerritMerged opendev/system-config master: reprepro : add labels
*** priteau has quit IRC12:48
openstackgerritAlbin Vass proposed zuul/zuul-jobs master: Roles to create, cleanup and promote snapshots in ec2
*** slaweq has quit IRC12:53
*** slaweq has joined #opendev12:53
*** priteau has joined #opendev12:54
*** slaweq has quit IRC12:56
*** slaweq has joined #opendev12:56
*** amoralej|lunch is now known as amoralej13:08
openstackgerritdaniel.pawlik proposed opendev/lodgeit master: Fixed issue related to newer Pillow lib
*** dhellmann has quit IRC13:19
*** dhellmann has joined #opendev13:25
*** vishalmanchanda has quit IRC13:26
*** ykarel_ has joined #opendev13:36
*** ykarel has quit IRC13:39
*** priteau has quit IRC13:43
*** mfixtex has joined #opendev13:43
*** priteau has joined #opendev13:46
*** rpittau is now known as rpittau|afk14:15
*** ykarel_ is now known as ykarel14:26
clarkbI've rechecked to keep moving on the inmotion mirror swap14:33
clarkbianw's investigation was unable to reproduce the failed selenium runs against review so that should pass14:34
clarkbif it doesn't I guess I hold the jobs on that change14:34
*** rsubrama has joined #opendev14:35
clarkbfrickler: fungi: re meetpad it seems that the experience depends a lot on the client side. Performance, browser, etc14:37
clarkbI think we still recommend chrom* and you may need to ensure you aren't trying to compile software or do other cpu intensive work at the same time14:37
clarkbif users start having problems then you can improve things by disabling videos14:37
clarkb(but as far as we've been able to tell the bottlenecks haven't been the server side as much as the client side)14:38
*** lbragstad_ has joined #opendev14:38
*** ysandeep is now known as ysandeep|away14:40
fricklerclarkb: can a single user disable receiving videos? or would every participant have to turn their camera off14:42
*** lbragstad has quit IRC14:42
clarkbthat is a good question. I don't know the answer to that14:43
clarkbI think the moderator can mute audio and turn off video for other participants globally within a call14:44
frickleralso I'm not sure whether that would really be an option for teams if they can work just fine from their pov with zoom14:44
fricklermaybe there is a server option to limit video bandwidth used by limiting resolution and/or frame rate?14:45
clarkbya I think this is a major reason why zoom doesn't use webrtc and ships a webasm bundle that does all the conferencing instead of webrtc in the browser14:45
clarkbfrickler: looks like you can limit video resolution in the config14:45
clarkbnot sure what that looks like with the upstream images, but jitsi meet appears to support it at least14:46
*** ykarel is now known as ykarel|away14:50
*** snapdeal has quit IRC14:52
*** mfixtex has quit IRC14:55
*** ykarel|away has quit IRC14:59
*** hashar has joined #opendev15:03
clarkbfungi: if you get a chance is one approach to fixing python3.5 testing for pbr15:07
clarkbI wonder if that breaks self testing in the unittests though15:07
clarkbit mustn't if we were relying on easy_install previously though15:08
clarkbthe --no-thin option is blocking on that15:08
clarkbanother option would be to stop testing 3.5 and just try to avoid adding specific >=3.6 behavior (made easier by testing 2.7)15:08
*** hashar has quit IRC15:10
*** hashar has joined #opendev15:10
*** hashar has quit IRC15:10
*** hashar_ has joined #opendev15:10
fungiclarkb: frickler: the lowest "performance" setting in jitsi-meet is audio-only sending/receiving, it's just not clearly labelled that way15:11
fungiindividual participants can set it15:11
openstackgerritAlbin Vass proposed zuul/zuul-jobs master: intercept-job -- self-service SSH access
fungiclarkb: on 787297 how is the requirements.txt addition working around things? is it because of the siblings install phase explicitly installing requirements without installing the project itself?15:19
clarkbfungi: it is because tox in general (including siblings) installs requirements before installing the project15:19
clarkbat least the way we typically use it15:19
fungigot it. so explicit project installation in a virtualenv on that platform would still fail, but tox is doing an end-run around setup_requires15:20
*** rsubrama has quit IRC15:29
clarkbfungi: actuall I'm going to WIP that change15:33
clarkbor no nevermind15:33
clarkbsuddenly I decided it really should be a test-requirement, but I don't think there is much harm in the change as is so nevermind15:33
clarkbbut if you think it should be a test-requirement I can move it15:34
*** amoralej is now known as amoralej|off15:37
fungiwell, is it also a runtime requirement?15:39
fungilooks like it's not15:39
fungiso yeah, maybe test-requirements.txt would make more sense15:39
fungiclarkb: thanks, i unapproved it for now15:40
clarkbok I'll update it momentarily15:41
openstackgerritMerged zuul/zuul-jobs master: intercept-job -- self-service SSH access
openstackgerritClark Boylan proposed opendev/git-review master: Install PBR explicitly
clarkbfungi: ^ that should still function but be a bit more clear15:43
fungicool, let's see if it still fixes the tests15:43
clarkbbtw not sure I'll get to the zk stuff today since I want to finish up the inmotion mirror things and clean up networking there since that is in a half done state15:45
clarkbmay be for the best to avoid having zk in a half cluster state over the weekend15:46
clarkbfungi: the gerrit testing has been looking happy though so I suspect we're still good ot do that gerrit update today15:46
clarkbthat will get us 3.2.8 as well as your jeepyb fixes15:46
*** mlavalle has joined #opendev15:47
*** hashar_ has quit IRC15:53
corvusi'd like to restart zuul today to pick up the global repo state patches15:55
corvuslast time we saw a behavior change with this (the fast-forward stuff); we think that's addressed15:55
clarkbcorvus: that requires a full restart too as it impacts the client side of the mergers/executors too?15:56
clarkb(it definitely requires the executors and mergers at least)15:56
corvusclarkb: yes -- but i've been doing full restarts recently anyway (since everything we've done recently touches the scheduler)15:56
corvusthings look generally quiet on the third column of the status page15:57
corvusso i think i'll do that now15:57
fungisounds good. thanks15:58
corvusran the pull playbook; restarting now15:59
fungiwe're also planning to do a gerrit restart later today15:59
fungionce ptg sessions are finished16:00
clarkbyup and I'm trying to get the new inmotion cloud reconfigured networking wise to get as many IPs out of it as we can16:01
corvus#status log restarted zuul on commit d4c7d293609a151f0c58b2d8a6fe4ac2817ee501 to pick up global repo state changes16:05
openstackstatuscorvus: finished logging16:05
clarkbjobs are running. Don't see any that have managed to finish already16:08
clarkbin another minute or two we should have some results though16:09
clarkbthe first failure appears to be a legit linting error so thats good16:12
clarkbwe have a number of successes now too16:12
corvusand failures;  a good healthy mix!16:13
corvusall done16:19
clarkbfungi: huh the py35 job failed with pbr in test-requirements16:22
clarkbI really didn't expect that /me looks more closely at the tox config16:22
corvusclarkb: that's pre zuul restart right?16:22
clarkbcorvus: no, after16:22
clarkbfwiw it seems to have checked out the correct commit based on the log so I don't think the failure is due to the restart16:24
clarkblikely the failure is due to the move from requirements.txt to test-requirements.txt and I don't undersatnd some subtle detail in why that makes a difference here16:24
corvuspassed for all the other vers16:24
clarkbya its a python3.5 specific issue we are trying to fix16:24
clarkbtl;dr is is requiring sni, for some reason xenial python 3.5 doesn't do sni even though claims it should16:25
clarkbthis causes easy_install to fail when trying to install setup_requires for git-review (specifically pbr)16:25
*** hamalq has joined #opendev16:27
clarkbfungi: it looks like tox runs sdist-make before installing deps. I'm beginning to wonder if the earlier successes were simply lucky and hit when they were not doing a brownout16:28
*** hamalq_ has joined #opendev16:33
openstackgerritClark Boylan proposed opendev/git-review master: Install PBR explicitly
clarkbmaybe that will work (using tox's skipsdist flag16:34
*** hamalq has quit IRC16:36
fungicould it be that we preinstall requirements.txt but maybe not test.requirements.txt?16:40
clarkbfungi: no, at least from local testing it is always the sdist-make step first then it installs requirements and test-requirements in one command16:41
clarkbI'm beginning to think the issue is the brownouts not being a 100% failure16:41
fungicould be, i'm also a fan of using skipsdist16:41
clarkbif I'm reading this correctly the only reason to not skipsdist is the ensure that you can make an sdist?16:45
clarkbit doesn't provide any functional benefit to the installation of the project later which can be done via what i think is the make a wheel and install that step16:46
fungithere was a time when we explicitly did sdist build in pep8 jobs to cover that case16:48
fungiheh, looks like we meant to do it in bindep:
fungityping is hard16:51
fungimeant to use skipsdist=true i mean16:51
clarkbI made the same typo when testing things for git-review not that long ago. noticed it didn't work and remembered the 's' at that point16:53
*** jpena is now known as jpena|off16:53
*** eolivare has quit IRC16:55
fungi#status log Removed address override on now that v6 routing between them works again16:58
openstackstatusfungi: finished logging16:58
*** andrewbonney has quit IRC17:00
openstackgerritJeremy Stanley proposed opendev/bindep master: Correctly skip sdist building in tox.ini
clarkbfungi: the failures now appear to be related to actual test failures17:10
clarkbmakes me wonder if not sdisting breaks the test suite some how17:11
clarkbwhich git-review is failing? so ya maybe17:12
clarkbin my local testing with py38 it installed git-review==2.0.1.dev317:13
clarkbbut in the test job it appears to have not installed git-review at all17:13
* clarkb git cleans and tries again17:15
clarkbhuh ya if I clean out .tox (I used -r anyway) and dist/ then I don't get a git-review installed17:17
openstackgerritClark Boylan proposed opendev/git-review master: Install PBR explicitly
clarkbthat installs . as a deps17:18
clarkbnot sure if that will work, but lets find out17:18
fungii use skipsdist in other projects and don't need to explicitly install "."17:22
fungialso usedevelop factors into that17:22
*** dtantsur is now known as dtantsur|afk17:22
fungii turn on usedevelop=True in the default testenv and then set it false in testenvs which don't need the project installed17:23
*** ralonsoh has quit IRC17:23
*** lbragstad_ is now known as lbragstad17:23
fungie.g. linters generally don't need the project package installed into the testenv17:24
clarkbah usedevelop would probably work for this, let me test that17:25
*** mfixtex has joined #opendev17:34
fungithough in my own projects i also have an actual [testenv:dist] where i explicitly `python bdist_wheel sdist` and then `twine check --strict {toxinidir}/dist/*`17:36
fungiso i'm directly testing dist creation, and that doubles as part of my release process too17:36
*** slaweq has quit IRC17:38
clarkbok usedevelop has to go in a testenv not a tox section17:42
clarkbthat took far too much time to understand17:42
openstackgerritClark Boylan proposed opendev/git-review master: Install PBR explicitly
clarkbI think ^ that should work17:42
hrw - thanks to ianw, kevinz, clarkb17:43
fungihrw: awesome!17:44
hrwfungi: I was surprised.17:45
hrwfungi: as all I did was just writing frustration and then Cc: proper people17:46
*** cloudnull has quit IRC18:11
openstackgerritMerged opendev/system-config master: Add mirror02 to inmotion
clarkbfungi: the latest ps on the git-review change passed on py3518:15
fungiright on, reapproving18:19
*** auristor has quit IRC18:40
*** auristor has joined #opendev18:43
clarkblooks like LE job on the new mirror failed again. This time I was careful to add the handler though :/ I'll see why that happened but then likely rerun the service-mirror.yaml playbook against it like I did with the last one18:51
clarkbit failed with The task includes an option with an undefined variable. The error was: 'ansible.vars.hostvars.HostVarsVars object' has no attribute 'letsencrypt_certcheck_domains' trying to update the domain list for certcheck. It also seems to have failed doing LE on nb03 which amybe triggered teh cacti issue for certcheck?18:53
clarkbrunning the mirror playbook now18:53
fungithat's a confusing error condition18:54
openstackgerritMerged opendev/git-review master: Install PBR explicitly
clarkbI've rechecked the --no-thin change18:56
fungilooks like we have 16GB free for the rootfs on review.o.o, and the docker image list has a opendevorg/gerrit 3.2 created 19 hours ago. current image we've been running is df1c7ac8dcae (no tag) from 2 months ago18:57
fungii did a `docker-compose pull` which doesn't seem to have updated the image list at all18:58
fungii guess we can restart the gerrit container at 19:00 utc18:58
clarkb57df55aec1eb is the new image18:58
clarkbdf1c7ac8dcae seems to be what we have been running on18:59
fungistatus notice The Gerrit service on is being restarted to pick up some updates, and should be available again momentarily18:59
fungidoes that look good?18:59
fungi#status notice The Gerrit service on is being restarted to pick up some updates, and should be available again momentarily19:00
openstackstatusfungi: sending notice19:00
-openstackstatus- NOTICE: The Gerrit service on is being restarted to pick up some updates, and should be available again momentarily19:00
fungiit's starting back up now19:00
clarkbsome complaints about not being able to compress jvm logs19:01
clarkband the plugin manager complaint that we expect19:01
fungiyeah, i've seen both on previous restarts19:01
clarkbseems to be an issue renaming the jvm logs from .log to .log.gz, weird19:02
clarkbbut also unlikely to be a significant issue19:02
fungimight be a file conflict19:02
clarkbI can load my dashboard and a change19:03
fungimaybe old compressed file with different permissions?19:03
clarkbah ya that could be19:03
fungiyeah, it's loading for me now too19:03
clarkbthe zuul summary still works19:03
clarkband the reported version is about what I expected so yay19:03
openstackstatusfungi: finished sending notice19:03
clarkb also looks good now after running service-mirror against it19:04
clarkbfungi: can you review as the next change in that series when you get a moment?19:04
* fungi just realized he wrote " instead of opendev in the status notice. oh well19:04
clarkbthen I can tail apache logs on the other mirror and wait for it to idle, delete it, then delete the network stuff we don't need anymore then bump the quota back to 819:05
clarkbfungi: oh I didn't catch it either, the .o.o's all run together sometimes19:05
*** yoctozepto has left #opendev19:09
*** yoctozepto has joined #opendev19:11
clarkbI guess I shouldn't delete the old host until the cleanup change lands to remove it from the inventory. But we should be able to safely approve the cleanup change, then wait for the server to idle before deleting it (since the mriror configs won't chagne in the interim)19:13
clarkbfungi: that means is also ready for review now19:13
fungialso gives nodes which were hitting the old mirror some time to notice the dns update19:13 liiiiiives!19:14
fungiin our memories19:14
fungiincluding my finger memory apparently19:14
clarkbfungi: skimming my todo list the other thing I have been trying to not let fall behind is the openstack project reparenting. Any idea if we are ready to move forward on those?19:15
clarkbI have suspect not as people were distracted by the ptg this week19:16
fungiwaiting to hear back from elod that things were good on the first batch we approved19:16
openstackgerritMerged opendev/ master: Flip mirror.iad3.inmotion to mirror02.iad3.inmotion
fungitesting of those has probably been delayed by ptg activity19:16
clarkbI'm tailing the mirror_443_access.log on each of the mirrors to get a sense for when load has shifted19:17
clarkbI wonder remove the old server until that happens and has merged19:17
openstackgerritMerged opendev/git-review master: Add option for disabling thin pushes
clarkbI wonder remove?19:18
clarkbI won't remove19:18
clarkbfungi: ^ want to do a git-review release on a friday ?19:18
clarkb--no-thin is the only consequential change (manpage was updated as was the pbr test fix)19:19
fungii suppose we can, or we could wait for monday19:21
*** Eighth_Doctor is now known as Conan_Kudo19:27
clarkbI'm not actually in a rush19:28
clarkbbut having it out before we next have to help troubleshoot that issue would be godo so sooner than later19:28
corvusi feel like git-review is an ok thing to release on fri19:28
*** Conan_Kudo is now known as Eighth_Doctor19:29
clarkbprobably is more urgent if I can get reviews on that to make the old mirror deletable19:29
clarkbI see both mirrors serving requests now fwiw19:29
clarkbneat I see requests to both from the same client IP at roughly the same times19:32
fungii can work up a git-review release in a few minutes19:36
clarkbthanks for the reviews!19:41
clarkbfungi: is expected to be sufficient for fixing the ansible facts issue on bullseye? your suggesting to install base-files from unstable implies no but ianw's email seems to imply that will make lsb_release work and then ansible will be happy?19:44
clarkboh wait I see its the chicken and egg issue between update and knowing what your repos are19:46
*** cloudnull has joined #opendev19:47
fungiyeah, the reprepro change was worth a shot19:51
fungiand arguably more correct than what we had19:51
clarkb is the other related fix which failed in testing (my comment calls out how). I'm not quite sure why the test failed though19:52
clarkbaha that also runs outside of the chroot19:53
openstackgerritClark Boylan proposed openstack/diskimage-builder master: debian-minimal: Set bullseye version
clarkblets see if that is happier19:54
*** stevebaker has quit IRC19:55
*** mfixtex has quit IRC20:23
openstackgerritMerged opendev/system-config master: Cleanup mirror01.iad3.inmotion
*** elod has quit IRC20:40
clarkbThe last request to ^ was almost an hour ago. I'll give it to the hour mark then proceed wtih cleanup20:40
*** elod has joined #opendev20:42
*** sboyron has quit IRC20:45
clarkbalright only new requests are from crawlers. I'm going to clean up the mirror now20:50
clarkbok mirror has been cleaned up as has the extra network/subnet/router that consumed and IP and the floating IP that the mirror was using. I've bumped the quota back up and the max-servers: 8 value should be used now20:58
*** whoami-rajat has quit IRC20:59
fungiclarkb: git-review 2.1.0 because --no-thin is technically a new feature, or 2.0.1 because it's an off-by-default bug fix/workaround?21:01
clarkbI think 2.1.0 is more correct21:01
fungii guess we did call it a feature in the release notes too21:02
fungicool, working on that now21:02
clarkbyup we should do 2.1.0 for sure given that21:04
fungiokay, tag pushed, i'll send out a release announcement in a bit once that's uploaded21:07
clarkbsounds good, thanks!21:08
clarkb pypi has the new release21:47
fungiyup, will write up something for service-announce shortly21:55
fungijuggling with evening chores21:55
zigoclarkb: fungi: Do you guys know why is failing?21:55
clarkbzigo: I don't know why the latest patchset is failing, but I think I figured out why the first patchset failed and that is why I pushed the new patchset21:56
zigoOh ok. :)21:56
fungialso we're setting the debian label on the mirrors in reprepro now21:57
zigofungi: So that should fix it if I retry tomorrow, with a new fresh image?21:57
clarkbzigo: I think the assumption is we need the dib hack21:58
clarkband seems unlikely we'll get that deployed by tomorrow21:58
fungiyeah, the one thing which dawned on me might solve the chicken-and-egg problem from the mirror end is that we build images using one of our mirrors21:59
fungiso it should be picking up that label at chroot build time now21:59
fungiso it might work with the next image update21:59
fungizigo: also, i don't suppose you have any idea how long it will take base-files 11.1 to migrate to bullseye?22:00
fungithat will make workarounds like 787665 moot, i think22:00
clarkbthe fedora-32 functioanl build job hit a nova api error trying to build an unmanaged VM outside of nodepool. I think we can recheck for this one. I'm checking the xenial job now22:00
zigofungi: Likely, it will migrate the day of the release, no?22:01
fungii honestly can't remember if it gets held until release or just into the freeze22:01
zigoOn the release list, they talked about release Bullseye in one week-end of may, but I'm not sure if a day is decided.22:01
fungiyeah, last i saw there still wasn't a day set22:02
openstackgerritClark Boylan proposed openstack/diskimage-builder master: Install pbr before glean to address SNI issues
clarkbzigo: ^ I think that is the fix for the bullseye fix22:05
clarkbI'm guessing that when pypi said only 1% of requests don't use SNI they didn't look at the nature of those requests. Likely all easy_install setup_requires requests22:06
fungiwhich is a vanishingly small number since newer setuptools calls out to pip22:07
openstackgerritClark Boylan proposed openstack/diskimage-builder master: debian-minimal: Set bullseye version
clarkband now rebased to be double sure22:07
* zigo crosses fingers22:07
*** slaweq has joined #opendev22:09
*** slaweq has quit IRC22:20
zigoLooking at the debian-release list, it looks like 22 of may could happen *IF* kibi can fix the debian-installer in time. If not, then June ...22:51
zigoThe major blocker being black screen on AMD GPUs ...22:51
openstackgerritMerged opendev/bindep master: Correctly skip sdist building in tox.ini
clarkbthe unexpected nova API error hit the centos8 job on but the xenial build there did pass23:04
clarkbI'm beginning to think we may need to grab nova logs from those jobs at least temporarily to debug furhter23:04
clarkbbtu I'm running out of day23:04
fungigit-review 2.1.0 release announcement is sent23:05
clarkbin any case looks like my fix did address the pbr thing properly since that job now passes (and it does a full boot test which will exercise the glean install)23:10
clarkbfungi: thanks23:10
ianwclarkb: thanks for looking at the xenial issue for dib23:58
*** tosky has quit IRC23:58
ianwUnexpected API Error. Please report this at and attach the Nova API log if possible.23:59
ianw<class 'neutronclient.common.exceptions.Unauthorized'> (HTTP 500) (Request-ID: req-2a6715af-ced6-4884-bd7c-18554d8b7947)23:59

Generated by 2.17.2 by Marius Gedminas - find it at!