[00:13] thumper: I figured out the blobstore thing. Although Juju doesn't use github.com/juju/blobstore any more, charmstore.v5 still does. I've emailed rog about it. [01:02] thumper: can you take another look at my PR? [01:42] thumper: u r an OCR \o/ my monday might get better after all :D could u PTAL http://reviews.vapour.ws/r/5590/? [01:44] yep [01:55] tyvm! [02:21] thumper: one line dependencies update - http://reviews.vapour.ws/r/5591/ [02:22] anastasiamac: timestamp is a lie [02:22] anastasiamac: do this: [02:22] godeps ./... | grep "juju/util" [02:22] you can then just copy that line [02:23] veebers: now good? [02:24] thumper: done \o/ [02:24] anastasiamac: I used to hand craft those lines until I worked out that trick [02:25] thumper: u r awesome \o/ we should have this in our wiki, I think :D [02:26] thumper: i need to update these licecne on 1.25 branch of juju/utils but no amount of fetching/pull is giving me this branch to creat my own on... [02:26] i suspect my upstrem/origin pair r not correct but can't figure out where m going wrong [02:26] could anyone point me in the right direction? [02:27] anastasiamac: probably not worth fixing IMO [02:28] there are hoops we could jump through if someone thinks it is important enough [02:28] bit kinda icky due to godeps [02:28] I know we were looking at updating godeps to allow it to look at different branches [02:28] but AFAIK, this isn't done yet [02:29] thumper: the thing is i don't think it's too hard.. i think the problem is my setup.. juju/utils does have 1.25 branch which juju 1.25 uses.. my thinking i branch from utils/1.25 to change licence and then do update deps for juju/1.25... [02:30] anastasiamac: well, it kinda is hard because godeps expects the revision to be in the master history [02:30] what you have to do is make a branch from the 1.25 hash [02:30] make the change [02:30] the merge in master [02:30] resolve conflicts [02:30] then propose that to merge into master [02:31] even though it may not change any files [02:31] thumper: is this not utils for 1.25? https://github.com/juju/utils/tree/1.25 [02:31] then you need to update the hash that 1.25 uses to be the non-mainline hash [02:31] that is the tag on master from which 1.25 was released [02:32] well branch [02:32] but you can't specify non-master branch in godeps [02:32] yet [02:32] ah ... k.. i'll mark as won't fix on 1.25 and if it makes someone cry, we'll fix then :D [02:32] it is not impossible [02:33] just difficult at the moment [02:34] of course \o/ but since 1.25 is in critical only, i can downgrade it to High and if there is a burning desire to have it in, then we'll go the painful way... [02:34] m sure there are other licencning inconsistencies in 1.25 deps :D [02:36] anastasiamac: menn0 thinks I'm wrong [02:36] which may well be trye [02:36] true [02:37] thumper: menn0: which part wrong? tracking only master branches in dependencies.tsv? [02:37] to create a branch off the 1.25 branch do this [02:37] git branch update-1.25-foo upstream/1.25 [02:37] then checkout [02:38] we were arguing about how git works when fetching revisions [02:43] thumper: menn0: fatal: Not a valid object name: 'upstream/1.25' [02:44] i think my remotes r funny [02:44] anastasiamac: do you have upstream set as the remote? [02:44] anastasiamac: what does "git remote -v" show? [02:44] $ git remote -v [02:44] origin https://github.com/anastasiamac/utils.git (fetch) [02:44] origin https://github.com/anastasiamac/utils.git (push) [02:44] upstream https://github.com/juju/utils (fetch) [02:44] upstream https://github.com/juju/utils (push) [02:45] menn0: ^^ [02:50] anastasiamac: that looks ok [02:50] anastasiamac: what about "git branch -r | grep 1.25" ? [02:51] menn0: and yet it does not work :( must b monday... 'git branch' yields nothing, just returns [02:52] anastasiamac: ok, do a "git fetch" and then try again [02:52] menn0: did and it just returns too [02:52] somehow m not "seeing" 1.25 in my git [02:52] but I can see it against juju/utils on github in browser [02:53] thumper: wallyworld: veebers: is it normal that landing jenkins says "(pending—juju-core-slave is offline)" [02:54] nope [02:55] anastasiamac: do you find that git asks for your github password everytime you push? [02:55] yes [02:55] anastasiamac: considered using ssh rather than https? [02:56] anastasiamac: I'll see if I can find out whats happening [02:56] anastasiamac: make sure github has your public key [02:56] i'd consider anything at this stage especially if it'll solve my current trip [02:56] eithet that or use a keyring [02:56] then set the remote url [02:56] git remote origin set-url git@github.com:anastasiamac/utils.git [02:56] then no more password requests [02:57] anastasiamac: do `git branch -r` [02:57] anastasiamac: also worth trying "git fetch -v upstream" to find out more about the fetch [02:58] thumper: $ git remote origin set-url git@github.com:anastasiamac/utils.git [02:58] error: Unknown subcommand: origin [02:58] anastasiamac: I got set-url and origin the wrong way around [02:59] menn0: with "git fetch -v upstream" i can se that 1.25 there but it's not being fetched for me :( [03:00] anastasiamac: I've kicked that node back into action [03:00] veebers: awesome [03:00] thumper: menno: finally after setting up ssh, i can checkout 1.25 \o/ [03:00] m moving on to creating my branch - THANK YOU :D [03:03] anastasiamac: schweet [04:04] thumper: prechecks wired up inside the InitiateMigration API handler: http://reviews.vapour.ws/r/5595/ [04:04] thumper: so much better from a UX perspective (see QA steps) [04:04] looking [05:32] thumper: just figured out the bug that jam filed a while back where migrations get stuck. it happens when you try to migrate a model back to a controller it's just been migrated from and the migrationmaster is still finishing up there. [05:33] thumper: the cause is subtle but I know how to fix it now. === frankban|afk is now known as frankban [07:43] axw: hey, hope your macaroon stuff is going ok. you probs don't have time today, but if you did get a few minutes free, would love a pretty straightforward review on a change to list-models output http://reviews.vapour.ws/r/5594 [07:44] wallyworld: going ok, but slowly. will take a look [07:45] only if you are able === admcleod_ is now known as admcleod [09:14] dimitern: ping - can we sync? [09:15] frobware: hey, sure, just give me 5m [09:16] dimitern: ok, will be in standup HO [10:25] wallyworld: Around? [10:44] babbageclunk: ping [10:44] voidspace: pong [11:14] I have a test failure on master [11:14] state/metrics_test.go:748: [11:14] c.Check(metricBatches[0].Unit(), gc.Equals, "metered/0") [11:21] godeps? [11:21] voidspace: ^^ [11:21] babbageclunk: hey [11:21] babbageclunk: nope, fully godep'ed up [11:21] wallyworld: hey! [11:21] babbageclunk: probably a map ordering issue [11:22] morning all btw [11:22] perrito666: o/ [11:23] wallyworld: Replying to your email - maybe worth a hangout if you can? (Sorry, realise it's late for you) [11:23] babbageclunk: no worries, give me 5 and i'll ping [11:23] wallyworld: awesome, thanks [11:29] babbageclunk: yeah, the result has two entries and they're swapped over [11:29] babbageclunk: so the test is order dependent [11:29] babbageclunk: stink [11:29] babbageclunk: easy enough to fix, I'll do it in my branch [11:33] dimitern: which HO? [11:33] frobware: I've realized it's not there so I just added one [11:33] https://hangouts.google.com/hangouts/_/canonical.com/juju-dns-nss [11:33] jam: ^^ [11:34] dimitern: brt [11:36] wow 500M of Res memory is a bit of a heavy footprint [11:45] Bug #1613992 changed: 1.25.6 "ERROR juju.worker.uniter.filter filter.go:137 tomb: dying"

[11:47] babbageclunk: free now, did you want to hang out? [11:48] wallyworld: yes please - how about in core? [11:48] sure got a link handy? [11:49] i don't have that one visible in my calendar [11:49] wallyworld: https://hangouts.google.com/hangouts/_/canonical.com/core [12:21] babbageclunk: ah damn, missed one place - ControllerInstances is called when restoring from backup [12:21] wallyworld: doh [12:21] so we'll need that query by tag [12:21] damn [12:21] Oh well [12:21] so close [12:22] But it's still only query by one tag for now, right? I don't need to do multiple calls and then intersect them together? [12:22] * perrito666 hears backup and peeks in === freyes__ is now known as freyes [12:52] wallyworld: Oh no, I do - the machines in the controller model are not all necessarily controllers. [12:53] correct [12:53] iscontroller=true && controller-uuid=blah [13:46] dimitern: I pushed an update for bridging all interfaces - https://github.com/frobware/juju/tree/master-bridge-all-interfaces [13:46] frobware: ok, looking [13:47] dimitern: that's just the unit test cases updated (and the bridge script updated). Was going to try some genuine deployments next. [13:47] frobware: at first glance looks like it does what we need [14:03] voidspace: standup? [14:03] frobware: oh yeah, thanks === petevg is now known as petevg_afk [14:48] frobware, voidspace, babbageclunk: can you please have a look at this tiny PR: http://reviews.vapour.ws/r/5597/ [14:56] dimitern: LGTM! [14:58] babbageclunk: thanks! [14:58] frobware: I'll land this unless you have comments I guess? [14:58] dimitern: was just looking at it [14:59] frobware: ok, no rush then [15:05] dimitern: do you know if the sysfs paths are consistent across kernel versions? [15:06] frobware: as far as I can see - yes, tested on trusty, xenial, and centos 7 [15:08] dimitern: I wonder why "brif" is not listed in here: https://www.kernel.org/doc/Documentation/ABI/testing/sysfs-class-net [15:11] frobware: not sure, but I've seen lots of examples using brif [15:13] dimitern: OK I see this in if_bridge.h - #define SYSFS_BRIDGE_PORT_SUBDIR "brif" [15:13] frobware: also this - https://books.google.bg/books?id=ALapr7CvAKkC&pg=PT420&lpg=PT420&dq=sys+class+net+bridge+brif&source=bl&ots=gVdzaVkYPv&sig=PVTMP6WszdmaCcAj4v6hNmYY67c&hl=en&sa=X&ved=0ahUKEwje3OPnwfjOAhUDWxQKHVJqBBYQ6AEIKTAC#v=onepage&q=sys%20class%20net%20bridge%20brif&f=false [15:13] dimitern: I'll believe the header file :) [15:16] dimitern: LGTM [15:16] frobware: :) thanks! [15:18] dimitern: stepping out for early dinner (read no breakfast and no lunch)... [15:19] frobware: enjoy! [15:19] dimitern: neither were intentional, just the way the day planned out === kadams54_ is now known as kadams54 === psivaa_ is now known as psivaa === plars_ is now known as plars === marlinc_ is now known as marlinc === zeestrat_ is now known as zeestrat === frankban is now known as frankban|afk [16:23] frobware: you've got a review on https://github.com/juju/juju/pull/6156 btw [16:37] dimitern: thanks [16:37] frobware: I'm about to propose the next step - decided to split networkingcommon changes in 2 PRs for easier review [16:37] dimitern: doh! :) [16:55] frobware: here's the next step: http://reviews.vapour.ws/r/5598/ [16:55] I don't expect you to review it tonight :) just FYI === akhavr1 is now known as akhavr [22:58] wallyworld: sorry I was a bit rushed in my review yesterday. I'll try again later. [22:58] wallyworld: it would be helpful if there were a definition for the access levels for models somewhere. we don't have that do we? [23:06] axw: there's a doc, let me find it, but it is a bit vague tbh [23:16] thumper: standup? [23:16] coming [23:42] thumper: have u made the decision about Jesse's MADE PRs? [23:45] no [23:45] I'm waiting until after 2.0 to look [23:45] no time right now [23:45] also [23:45] * thumper weeps at apiserver/params importing mongo [23:46] * thumper looks at perrito666 [23:47] UpgradeMongoParams [23:47] apiserver/params shouldn't import external packages [23:47] especially not for arg structures [23:47] I' [23:47] I'll fix it in my branch [23:47] because I have introduced the cycle