[00:30] babbageclunk, timClicks: vsan summary? [00:30] thumper: summary is: good - just testing out the non-hacked together version [00:31] cool cool [03:56] thumper: new azure regions turned out to be lots-o-fun https://github.com/juju/juju/pull/10518 [04:07] wallyworld: thanks for wrangling through all of that [04:08] babbageclunk: hey btw could you raise a PR against my branch? I think that thin disk provisioning may be closer than we think [04:08] but I need the VM template stuff working [04:15] timClicks: pushing as we speak [04:24] timClicks: https://github.com/timClicks/juju/pull/1 [04:25] ha, it's prompting me to delete my fork of timClicks/juju [04:26] and I'm all like "nuh-uh! If anything timclicks/juju is a fork of babbageclunk/juju!" [04:27] babbageclunk: thanks for ruining my day by destroying all of my stuff github [06:07] kelvinli_: i just need a +1 on a forward port of 2.6 https://github.com/juju/juju/pull/10519 [06:08] wallyworld: done! [06:08] ty [06:08] np [10:07] merge 2.6 branch of pylibjuju into master https://github.com/juju/python-libjuju/pull/338 - only 120k of changes :| [13:28] achilleasa, if we get rick_h to make a 2.7 branch once i've merged https://github.com/juju/python-libjuju/pull/338 into master, then you should just be able to copy the schema to the client folder and ln -s correctly with latest and then run make client [13:29] stickupkid: thanks for the tip! [13:29] stickupkid: land your branch first so the branch has that in it [13:29] rick_h, achilleasa yeah, i've kicked it off [13:29] achilleasa, pointers here https://discourse.jujucharms.com/t/python-libjuju/1553 [13:29] love me some discourse :D [13:32] rick_h, achilleasa it's landed [13:32] stickupkid: k, after the call I'll try to get it up [13:40] stickupkid: achilleasa created the 2.7 branch [14:17] rick_h, pylibjuju new version with CMR, 0.11.8 or 0.12.0 or 1.0.0? [14:18] stickupkid: hmm, we talked about matching juju right? [14:18] stickupkid: what do you think of the 2.6.x? [14:18] rick_h, well, yeah, that could work, as it would make sense [14:18] rick_h, should i mention anything in the readme about that? [14:20] stickupkid: yea, and maybe discourse about moving to matching the juju versions [14:20] stickupkid: always better to overshare vs under [14:20] rick_h, 100% agree [14:29] stickupkid: so for the bundle bit do I need to set my version to 2.7.0? 2.7-rc1 ? [14:30] achilleasa, i'm never sure, we shouldn't release it with that, so maybe rc1 to ensure people delete it and replace it with 2.7.0 [14:30] 2.7-deleteme [14:30] or 2.7-dev? [14:31] I see some with the -rcX suffix [14:35] 2.7-dev [14:35] achilleasa, the rcX suffix ones need to be removed [14:36] stickupkid: HO? [15:09] CR for 2.6.6 release for pylibjuju https://github.com/juju/python-libjuju/pull/339 [15:13] hml: I am about to finish up, but could you look at https://github.com/juju/juju/pull/10520 in your day? [15:14] manadart: sure. trade you for a short one? https://github.com/juju/description/pull/59 [15:14] hml: Done deal. [15:51] in an overlay, i needed to include a machine definition not used in the overlay but that *is* used in the accompanying bundle. what is the rationale there? [15:52] pmatulis, achilleasa might know that [15:53] pmatulis: is this with 2.6 or develop? can you share the bundle/overlay so I can take a look? [15:56] pmatulis: I think it's just a limitation of the merging logic [15:58] achilleasa, v.2.6.6 [15:58] rick_h, seems odd. especially when the error references the bundle and not the overlay [15:59] pmatulis: right, a matter of the merging/resolving process is my guess [15:59] pmatulis: can you please try the same with develop? We have completely refactored the merge logic for the upcoming 2.7 so it might work as expected with the latest edge [16:04] achilleasa, ohh, fyi: https://pastebin.canonical.com/p/CGPR6ZsnMF/ [16:07] achilleasa, will the upcoming changes have potential for breakage if people don't make changes? [16:08] pmatulis: no, the changes should be backwards-compatible (actually even better as some stuff like for instance the 'trust' flag cannot be currently overwritten via an overlay). The 2.7 version additionally supports the cmr-in-bundle work [16:11] stickupkid: one question on the versioning in the PR [16:12] rick_h, shoot [16:12] rick_h, actually i'll trade you a PR for a question https://github.com/juju/cmd/pull/66 [16:12] stickupkid: sorry, replied with it. mainly...as much as I don't want to...2.6.6.1? [16:12] hmmm [16:12] 2.6.6.0? [16:12] stickupkid: e.g. if we have to rev a libjuju version without a new juju version? [16:13] rick_h, yeah, but start a 0 [16:13] ? [16:13] stickupkid: is it enough to do 2.6 and then rev libjuju indepdenent? [16:13] stickupkid: fair enough [16:13] achilleasa, sweet, thanks [16:13] rick_h, "is it enough to do 2.6 and then rev libjuju indepdenent" - not sure i understand [16:13] pmatulis: the QA steps for https://github.com/juju/juju/pull/10476 include examples of the upcoming overlay changes. Note that the multi-doc yaml bundles are only processed by the new code path (but they are only used by the CMR features which older clients ignore anyway) [16:14] stickupkid: so do we say that python-libjuju tracks 2.6 and so 2.6.3 might actually have the release note item of "updated to 2.6.7 api changes" [16:14] stickupkid: e.g. since the api facades are locked in a way you should be able to use python-libjuju 2.6.x with any 2.6 juju? [16:14] rick_h, ah in that case shouldn't it be 2.6.7 and we jump versions? [16:15] rick_h, that might luck ugly though [16:15] * rick_h doesn't follow that... [16:15] rick_h, ho? [16:16] stickupkid: omw [16:31] first time i've seen one of my machines' state suddenly go from 'started' to 'down'. why does this happen? [16:31] machine agent croaks? [16:33] corresponding MAAS node shows it's powered off [17:00] pylibjuju 2.6.0 released with CMR support [17:00] manadart, released, i'll make a discourse post for it [17:06] manadart, here you go https://discourse.jujucharms.com/t/pylibjuju-2-6-0-release-notes/1926 [17:30] pmatulis: so the agent restarts as part of going up [17:30] pmatulis: if you time it right you can catch it in a down state but it should flip back [17:30] pmatulis: if it doesn't then something went boom/etc [17:31] rick_h, yeah, the node just crapped out on me. i had to power it back on via the MAAS web UI. everything looks fine now though. scary [17:31] pmatulis: ah, yea don't trust hardware :) [17:36] * rick_h secretly adds cards and sticks stickupkid's head on them for later in the week bwuhahahaha [18:44] hml: can I bug you to make sure I typed "python" right please? https://github.com/CanonicalLtd/juju-qa-jenkins/pull/261 [18:44] rick_h: sure, looking [18:48] rick_h: approved [18:51] hml: ty [20:13] rick_h, :worried: [20:13] stickupkid: lol, all good you already did it [20:13] just didn't see it right away [20:13] rick_h, haha === hml_ is now known as hml === narindergupta is now known as narinderguptamac [21:27] timClicks: any luck with the thin disk spike? [21:28] wallyworld: I've found the two places that we need to change, but something weird happened when I tried to bootstrap [21:28] the process hung indefinitely [21:29] could be related to the changes i guess. sounds like it needs a little more work to be acheievable [21:29] thanks for trying === narindergupta is now known as narinderguptamac [21:44] wallyworld: out of curiousity, how does juju/mutex work? I think bootstrap is hanging because a previous mutex hasn't been released [22:04] timClicks: it's a machine wide mutex which uses flock on *nix and something else on windows [23:34] review for someone: https://github.com/juju/schema/pull/19 [23:35] ugh, no merge bot set up for schema package [23:36] looks like we have never had one set up [23:37] babbageclunk: got a moment? ^^ [23:38] * thumper remembers that babbageclunk is busy this morning [23:38] wallyworld: ^^ [23:38] thumper: sure - in 1:1? [23:38] oh... you are here [23:38] babbageclunk: I was pointing to the simple schema review [23:38] I tried to use a schema.TimeDuration in controller config [23:38] and the error message it gave was not overly helpful, [23:38] so fixing the root problem [23:38] oh sorry, thought you were asking about a merge [23:39] -bot [23:39] ok taking a look [23:39] nah, just the PR [23:39] thumper: LGTM [23:39] the duration was added by copying the time one [23:40] unfortunately, I wrote the time one four years ago [23:40] so I can't even really blame someone else for missing it [23:40] thanks hpidcock [23:40] I also have approved it [23:41] * thumper waits just in case babbageclunk has a question [23:41] thanks [23:41] just in case [23:41] all good [23:41] only comment was about the horrible formatting you fixed [23:41] you mean gofmt fixed [23:41] Sure, your robot minion [23:42] sometimes I deliberately format code wrong so I can watch it fix it [23:42] now I need to remember how to update the vendor shit [23:42] make rebuild-dependencies? [23:42] anastasiamac: that is a bit of a burn [23:42] Oh but you need to add it to the toml file first [23:42] yeah [23:43] thumper: https://discourse.jujucharms.com/t/juju-dependencies-managed-by-dep/139 [23:43] thumper: haha u picked on that? :D [23:44] anastasiamac: thanks, I think I remembered enough [23:44] \o/ [23:44] that discourse post has instructions on how to bring/update deps for juju/juju [23:44] i refer to it when needed - less to remember :D [23:49] * thumper runs to the gym [23:56] wallyworld: PTAL https://github.com/juju/juju/pull/10523 [23:56] ok