[00:01] <thumper> timClicks: now?
[00:02] <timClicks> sure
[00:02] <timClicks> thumper: see you in 1:!
[00:07] <tlm> babbageclunk: got 5 minutes after standup? Need to pick your brain about manifolds
[00:08] <babbageclunk> tlm: yup yup!
[00:12] <wallyworld> thumper: a multiwatcher race fix https://github.com/juju/juju/pull/11223
[00:12] <thumper> will looks
[01:29] <wallyworld> thumper: ah, i just pushed a different change before i saw your comment
[01:29] <wallyworld> i can revert to what i had originally, or keep what i have now
[01:29] <wallyworld> my original change is safer because it handles all callers
[01:30] <wallyworld> whereas now in the second attempt, it only fixes just one caller and other can still do the wrong thing
[01:46] <thumper> wallyworld: I looked at the call site, and there are potentially long blocking things there
[01:46] <thumper> so decided that your original approach was good
[01:47] <thumper> babbageclunk: is the release going smoothly?
[02:09] <babbageclunk> thumper: huh - apparently not :(
[02:09] <babbageclunk> https://jenkins.juju.canonical.com/job/release-juju-stable/114/
[02:10] <thumper> :(
[02:11] <babbageclunk> digging now
[02:30] <babbageclunk> I don't get it - wait-for-operation finished but the proposed streams json file doesn't contain entries for 2.7.2
[02:30] <thumper> https://github.com/juju/juju/pull/11224 forward port of older PR
[02:34] <babbageclunk> Ok, I can see in the update-streams log that it's written a log line for this request
[02:35] <babbageclunk> but the json file says updated on 17th of Jan
[02:35] <babbageclunk> https://streams.canonical.com/juju/tools/streams/v1/com.ubuntu.juju-proposed-tools.json
[02:36] <babbageclunk> http://streams.canonical.com/juju/new-tools/juju-dist/tools/update-streams.log
[02:36] <thumper> is it putting in proposed or released?
[02:37] <babbageclunk> at this point in proposed
[03:11] <babbageclunk> ok, digging around in the code isn't really helping - rubber ducky?
[03:32] <wallyworld> babbageclunk: it looks like the job was run with agent_stream=proposed
[03:33] <wallyworld> surely it should be released?
[03:33] <babbageclunk> yeah, I think that's right at this point isn't it?
[03:33] <wallyworld> no
[03:33] <babbageclunk> I think it tests them and then updates to released
[03:33] <wallyworld> we are updating released streams for a released build
[03:33] <wallyworld> really?
[03:33] <babbageclunk> looking at the jobs again...
[03:34] <wallyworld> i could very well be wrong
[03:34] <babbageclunk> from this list of jobs https://jenkins.juju.canonical.com/job/release-juju-stable/
[03:35] <babbageclunk> It looks like it releases them to proposed and then to released.
[03:35] <wallyworld> ok, that i didn't realise
[03:36] <babbageclunk> And it's the first bit that doesn't seem to have worked - the s3 stuff has 2.7.2 and a recent update time, but streams.canonical.com doesn't.
[03:36] <babbageclunk> I'm going to increment the poke value and resume the build, maybe it was an eventual-consistency thing?
[03:37] <wallyworld> tlm: did you run into this issue last release? can you recall what was done?
[03:38] <tlm> wallyworld: catching up. We did have issues with this. It does go proposed => released from memory
[03:38] <wallyworld> yeah, i was totaslly wrong
[03:38] <tlm> I think I was updating the poke value
[03:38] <wallyworld> but it's not even going to proposed atm
[03:38] <babbageclunk> yeah, I'm rerunning it with poke=1 now
[03:39] <tlm> I think jerff gave me issues on release and simon had to get someone to do something on some server somewhere but I could be very wrong as well
[03:43] <wallyworld> hpidcock: the default task status is Completed. did you specify Running and still not get results?
[03:43] <wallyworld> i can double check to be sure
[03:43] <hpidcock> oh right. wallyworld I was just expecting `juju operations` to show all tasks
[03:44] <wallyworld> the actions api (currently unused but there) defaulted to Completed so i was just mirroring that
[03:45] <wallyworld> i think the idea was the people would usually want to see stuff that had finished
[03:45] <wallyworld> but i'm guessing
[03:45] <hpidcock> i think we should assume all
[03:46] <wallyworld> we can do that and ask for feedback
[03:46] <wallyworld> the original api is 5 years old and never really used
[03:46] <hpidcock> yeah lets go with what feels right and nice
[03:47] <hpidcock> like kubectl doesn't just show pods that are running when you go kubectl get pods
[03:47] <wallyworld> i was debating which was better but i think it's less of a surprise to go with all
[03:47] <wallyworld> get pods vs list a large set of things is different though
[03:48] <hpidcock> you can have like a 1000 pods in a k8s cluster
[03:48] <hpidcock> operations are also an aggregate
[03:48] <wallyworld> especially with all the extra juju operators :-)
[03:49] <hpidcock> I think we need some garbage collection on the list
[03:49] <wallyworld> we have it
[03:49] <wallyworld> there's a pruner worker
[03:49] <hpidcock> well then that sounds like the list will be somewhat small
[03:49] <wallyworld> depends - prune > 3 days old i think
[03:49] <hpidcock> it's also a model based command right?
[03:49] <wallyworld> yep
[03:50] <wallyworld> i added the batching stuff jus in case
[03:50] <wallyworld> we can always wrire that up if needed
[08:33] <zeestrat> Got a couple of PR's for Ansible in charm helpers if anyone could help with review and merge: https://github.com/juju/charm-helpers/pull/403, https://github.com/juju/charm-helpers/pull/404
[09:01] <nammn_de> manadart: if you find some time later could you take a look and see whether those API params and result changes in params/network.go makes sense? https://github.com/juju/juju/pull/11213/files#diff-b8ec81ce097ebb5fa3ef9f36b26bd128R934 With this I could start working on the cmd patch tomorrow or later after ci
[09:06] <manadart> nammn_de: Yes; though it looks like I will be working the release today.
[09:07] <nammn_de> manadart: sure no worries
[20:44] <anastasiamac> babbageclunk: want to catch up at some point?
[20:44] <babbageclunk> anastasiamac: yup yup
[20:45] <babbageclunk> might need to be getting the release sorted first though :(
[20:45] <anastasiamac> babbageclunk: awesome \o/ ping when u r free
[21:16] <evhan> I can't seem to find docs for the right form for "serviceAccounts" in the podspec now.
[21:17] <evhan> I know it *was* "serviceAccount", and I thought I then had it working with an array under "serviceAccounts", but that doesn't seem to be right (anymore?).
[21:23] <evhan> i.e. following this structure: https://discourse.jujucharms.com/t/updated-podspec-yaml-new-features/2124
[21:23] <evhan> (Under "Workload permissions and capabilities".)
[22:20] <wallyworld> evhan: from memory, the core yaml has serviceAccount (singular) which is just for the main workload pod. the additional k8s specific yaml has serviceAccounts (plural) to allow additional, arbitrary onces to be created
[22:24] <wallyworld> eg https://pastebin.ubuntu.com/p/FC3hDhJbzt/
[22:25] <wallyworld> kelvinliu: i think your vsphere pr is good to land? we can then do a 2.7.3 candidate snap
[22:27] <anastasiamac> wallyworld: btw ... ░H░A░P░P░Y░ B░I░R░T░H░D░A░Y░░♪
[22:27] <wallyworld> i was hoping to keep it quiet
[22:27] <tlm> happy birthday wallyworld
[22:27] <wallyworld> now the cat is out of the bag
[22:30] <anastasiamac> haha can't keep it quiet! if not be loud now, then when? :D
[22:30] <kelvinliu> wallyworld: yep
[22:31] <wallyworld> at my funeral
[22:31] <kelvinliu> wallyworld: happy birthday! 🎂
[22:32] <wallyworld> ty
[22:47] <babbageclunk> yay happy borthday wallyworld! Let's all sing at standup!
[22:51] <evhan> wallyworld: Yeah, I had that but get json: unknown field "serviceAccount" (from operator master).
[22:51] <evhan> Also, happy birthday. :)
[22:54] <evhan> Hmmm, that's what the provider expects too (ref. https://github.com/juju/juju/blob/develop/caas/kubernetes/provider/specs/v2_test.go). Maybe I'm doing something dumb here.
[23:10] <wallyworld> evhan: sorry, was in meeting. can you share the yaml you are trying? kelvinliu can help as well
[23:11] <kelvinliu> wallyworld: sure
[23:12] <kelvinliu> evhan: the top level is serviceAccount, but the one under kubernetesResources is serviceAccounts (because it's an array)
[23:12] <evhan> https://paste.ubuntu.com/p/W5Fpj4NqMk/
[23:12] <evhan> I noticed the model was on 2.7, so I'm just trying to upgrade and go again on 2.8-beta1.
[23:13] <kelvinliu> evhan: u have to use `version: 2` k8s spec
[23:13] <evhan> Hrnnnnn I've run into this before too
[23:13] <evhan> Thank you.
[23:14] <kelvinliu> the legacy doesn't support rbac
[23:14] <kelvinliu> np
[23:14] <evhan> Right, I left that key behind when moving values around. Sorry for the noise.
[23:17] <wallyworld> evhan: there's some doc improvements we are woring on. everything is a bit scattered on discourse
[23:18] <timClicks> "a bit"
[23:18] <wallyworld> well i had all the main doc posts pinned but they got unpinned :-(
[23:19] <timClicks> imo a good intermediate step would be edit cory john's post and remove all of the other pins
[23:20] <timClicks> because 9 pinned posts isn't effective either
[23:21] <evhan> FWIW when looking for help w/ k8s configs, all roads tend to lead to the "Updated pod spec yaml" post above.
[23:21] <evhan> Via search and links from other posts.
[23:21] <evhan> https://discourse.jujucharms.com/t/updated-podspec-yaml-new-features/2124 <----- that one
[23:25] <wallyworld> getting stuff off discourse and properly into docs is te best solution IMO. discourse posts is a terrible medium for docs
[23:26]  * timClicks nods
[23:26] <timClicks> ^ emphasis on the "intermediate"
[23:40] <hpidcock> kelvinliu: can you please do a 2.7 -> develop merge, there is a lot of k8s changes from you that need merging
[23:49] <anastasiamac> babbageclunk: about NewRestroreCommand and export_test... m k to pass in a connect func but it'll need to b accessible for main.go too... so
[23:49] <anastasiamac> babbageclunk: r u k for me to have connect exported from a database?
[23:55] <kelvinliu> hpidcock: sure
[23:55] <hpidcock> kelvinliu: thank-you
[23:55] <kelvinliu> np
[23:59] <babbageclunk> anastasiamac: yeah, I think that's fine
[23:59] <anastasiamac> k