| veebers | wallyworld: sweet, test highlighted the issue, quick fix in the end (conditional at wrong level, let probablySetStatusHistory make that choice) | 00:11 |
|---|---|---|
| wallyworld | yup, it checks for status being the same | 00:11 |
| veebers | aye, I had to tweak that a little with the history overwrite bits | 00:11 |
| wallyworld | kelvinliu_: no rush, just sometime today if possible, i added placement support https://github.com/juju/juju/pull/9190 | 00:59 |
| kelvinliu_ | wallyworld, yeah, will take a look soon | 01:01 |
| * wallyworld pops out to buy coffee | 01:02 | |
| veebers | wallyworld: seeing this in the operator pod log: https://paste.ubuntu.com/p/WS8STGfb9F/ did something land recently that might cause that? I'm pretty certain it's unrelated to my changes at all | 02:38 |
| wallyworld | operator pod code has changed, but your local copy of the code is stale | 02:39 |
| wallyworld | you just need to rebase | 02:39 |
| veebers | ack thanks, will do | 02:40 |
| veebers | wallyworld: ok, finally got that PR updated, history looks good: https://paste.ubuntu.com/p/FnVQ7q3QX8/ I did see these logs in the controlelr debug-log, I don't think it's my changes, I did touch the provisioner though: https://paste.ubuntu.com/p/pjfvpmYngV/ | 03:55 |
| wallyworld | veebers: yeah, those are "normal". just getting a coffee and then will look | 03:56 |
| veebers | ack, thanks. should be pretty much there this time :-) | 03:57 |
| * thumper sighs | 04:17 | |
| thumper | anastasiamac, wallyworld: do you know of any CLI tests that ensure there is a controller entry for a mock CLI test? | 04:18 |
| thumper | using testing.BaseSuite | 04:18 |
| thumper | which has a fake xdg data suite thingy | 04:18 |
| wallyworld | you mean in the local yaml files? | 04:18 |
| thumper | yeah, to fake a controller entry | 04:19 |
| wallyworld | hmmm, there's a few slightly different ways tests do it; i can't recall a specific example without digging around, i can take a look | 04:19 |
| thumper | oh... | 04:19 |
| thumper | it uses a different base... | 04:20 |
| thumper | uses JujuOSEnvSuite | 04:20 |
| wallyworld | there's a MemStore that tests can use | 04:20 |
| wallyworld | and that's passed ion to the CLI constrcution | 04:20 |
| wallyworld | and the mem store is set up with data | 04:20 |
| wallyworld | so the test never hits disk at all | 04:20 |
| thumper | is that in the model wrapper? | 04:21 |
| thumper | wallyworld: https://github.com/juju/juju/pull/9191 | 06:02 |
| wallyworld | righto | 06:02 |
| thumper | wallyworld: it seems github diff doesn't like me renaming the state_test.go to state_internal_test.go | 06:02 |
| thumper | and then add a different state_test.go | 06:02 |
| wallyworld | sigh | 06:02 |
| thumper | if only it tracked renames :) | 06:02 |
| * thumper EODs | 06:04 | |
| thumper | see y'all tomorrow | 06:04 |
| wallyworld | oh joy! +5,633 −5,423 | 06:04 |
| babbageclunk | wallyworld: yeah, that description seems too brief given the size of the PR | 06:13 |
| babbageclunk | wallyworld: want a shorter one? https://github.com/juju/juju/pull/9192 | 06:14 |
| anastasiamac | most of it just a test move, like file rename kind of... | 06:14 |
| babbageclunk | anastasiamac: ah right - yeah, there are a lot of status tests. | 06:14 |
| anastasiamac | yep | 06:14 |
| stickupkid | I loke how we have two different places for escaping mongo "." in names | 09:54 |
| stickupkid | s/loke/like | 09:54 |
| manadart | stickupkid: Regarding recent changes to CI for formatting, what might cause this? http://ci.jujucharms.com/job/github-check-merge-juju/3433/console | 10:25 |
| stickupkid | manadart: what go version you using? | 10:26 |
| manadart | gsamfira: Want to jump in here? stickupkid: should that matter for CI? | 10:27 |
| stickupkid | manadart: if it's 1.11, then you need to downgrade to 1.10.x - golang made a breaking change | 10:27 |
| stickupkid | RE: https://github.com/golang/go/issues/26098#issuecomment-400859370 | 10:28 |
| stickupkid | manadart: we're going to force 1.11 next week at the sprint, so all outstanding branches etc will need updating | 10:28 |
| gsamfira | stickupkid: gotcha. Downgrading. I'm running 1.11 | 10:30 |
| gsamfira | thanks | 10:30 |
| stickupkid | gsamfira: np :D | 10:32 |
| stickupkid | manadart: i'm looking at escaping documents from mongo, have you seen this in passing? or do we do it late in the day, before we use it | 11:12 |
| stickupkid | s/escape/unescape | 11:12 |
| stickupkid | mandart: I'm looking if there is any common pattern for this | 11:13 |
| manadart | stickupkid: Not sure. I've not had accommodate it directly. | 11:13 |
| stickupkid | manadart: fair | 11:13 |
| rick_h_ | stickupkid: manadart I thuoght we had a tool around that. I remember jam mentioning making things "fat hyphen" and such | 12:51 |
| stickupkid | rick_h: we do, i found it | 12:51 |
| stickupkid | rick_h: well i found it twice, there is code duplication :D | 12:52 |
| rick_h_ | stickupkid: hah, ok cool | 12:54 |
| hml | stickupkid: changes made: https://github.com/juju/charm/pull/264 | 14:09 |
| hml | stickupkid: i plan to squash the commits before merge. | 14:09 |
| stickupkid | hml: yeah, fine by me | 14:09 |
| stickupkid | hml: LGTM - :D | 14:09 |
| hml | stickupkid: merged | 14:13 |
| parlos | Good Morning | 14:57 |
| rick_h_ | morning parlos | 15:02 |
| rick_h_ | externalreality: QA'd the branch there. Looks awesome! | 15:02 |
| rick_h_ | externalreality: one typo note and I thought I recalled a conversation around the error at the end of complete? If that's for followup work then carry on | 15:02 |
| parlos | I've used the openstack-bundle-57 to deploy OS, works nice. However, I ran in to trouble a few miles down the line. What I discovered was that nova compute ran out of local storage, however afaik, there is no config option for that in the nova-compute service, as that is handed by Libvirt. Which then got me wondering, who installes/configures libvirt? Guess it comes with the OS, and if so, how could this be modified from the bundle? | 15:11 |
| stickupkid | hml: are you using "gopkg.in/charm.v6-lxdprofile" in the dep file | 15:12 |
| stickupkid | hml: i'm trying to do it the standard way, rather than injecting the git repo directly into the vendor folder... | 15:12 |
| rick_h_ | parlos: so the bundle is a collection so applications and each application will have its own config available. I'd expect that it's something you can tweak in the nova-compute charm config: https://jujucharms.com/nova-compute/ (look at the config section) | 15:17 |
| rick_h_ | parlos: for expertise on setting that up better I'd reach out to the openstack folks but hopefully that helps | 15:17 |
| parlos | rick_h; from that you can config the empheral storage, but the storage i think i'm refering too is managed by libvirt. which isnt a service in the bundle... | 15:19 |
| parlos | rick_h; from that I can config the ephemeral storage, but the storage i think i'm referring too is managed by libvirt. which isnt a service in the bundle... | 15:20 |
| parlos | rick_h; the question can be generalized, who configures the 'base' services on a node? SSH seems to be configured by maas in my case, but juju also does something too it (i think) but ssh is not listed as a service..hope my question is understandable. | 15:23 |
| stickupkid | hml: this is really annoying, we can't namespace dep toml file to point to "gopkg.in/juju/charm.v6-lxdprofile" | 15:26 |
| stickupkid | hml: it's because https://github.com/niemeyer/gopkg/blob/master/version.go#L16 hardcodes unstable :| | 15:26 |
| hml | stickupkid: poopy | 15:26 |
| stickupkid | hml: i've got an idea :D | 15:27 |
| stickupkid | hml: yes it works :D | 15:28 |
| hml | stickupkid: w00t | 15:29 |
| stickupkid | hml: one sec, i'll make a branch, that we can use as our base branch | 15:29 |
| stickupkid | hml: this could be away to get rid of the patches folder as well | 15:30 |
| hml | stickupkid: cool, jam was wondering if that could be done with deps | 15:31 |
| stickupkid | hml: yeap :p | 15:31 |
| stickupkid | hml: CR here https://github.com/juju/juju/pull/9193 | 15:32 |
| stickupkid | hml: so we don't have to change any source code in juju, just update the revision, when the v6-lxdprofile feature branch is done, then we just update the dep Gopkg.toml + lock file and we're done :D | 15:33 |
| stickupkid | hml: right, i can start working on the validation stuff now :D | 15:34 |
| stickupkid | hml: don't feel dirty hacking on my vendor folder now | 15:34 |
| hml | stickupkid: how does that work? charm/lxdprofile instead of charm/v6-lxdprofile? | 15:35 |
| stickupkid | hml: typo, just needs to be charm, one sec | 15:35 |
| stickupkid | hml: it ignores everything after the project anyway | 15:36 |
| stickupkid | hml: try now | 15:38 |
| hml | stickupkid: trying | 15:41 |
| rick_h_ | parlos: so juju will touch things on a node that it needs to communicate/work like ssh keys. It doesn't do anything else. All of that is up to the charms on it. So I'd expect that if nova-compute needs to manage things for vms it would have those knobs. Otherwise i'd expect some other charm to provide the tools/controls for something else running. | 15:41 |
| stickupkid | manadart: is there away to have multiple leases at once, i.e. could leases overlap and how do know which leases are being offered | 15:42 |
| hml | stickupkid: worked for me. :-) | 15:50 |
| hml | stickupkid: looking at the safeLXDProfile now | 15:52 |
| stickupkid | hml: https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcRIOg8AK-PpstPTj5MeHL-loVkvw4V9RRhFrwMs9sVnz5AP7-k1 | 15:52 |
| hml | :-D | 15:52 |
| parlos | rick_h: so, your guess would be that nova-compute (or some of the charms in the bundle) would install and configure libvirt in this case. | 15:54 |
| parlos | guess I have to do some digging, to see if thats something I can configure. thanks for the answers | 15:56 |
| rick_h_ | parlos: correct, that would be my assumption if nova is creating VMs on top of libvirt. That's its job to manage then. | 15:57 |
| stickupkid | hml: we should be able to do that from the charmstore now - in theory "juju deploy cs:~simonrichardson/lxd-profile-3" | 16:01 |
| hml | stickupkid: yes | 16:02 |
| hml | stickupkid: or from juju/juju : juju deploy --debug ./testcharms/charm-repo/quantal/lxd-profile | 16:02 |
| stickupkid | hml: i wanted to make sure that it worked from the charmstore :D | 16:03 |
| hml | stickupkid: did you push a charm up to the store? I haven't | 16:03 |
| hml | just been using local charms todate | 16:03 |
| stickupkid | hml: yeah, we can always remove it later, but will help with manual testing | 16:03 |
| hml | stickupkid: i believe we were going to add as juju-qa or something | 16:04 |
| hml | stickupkid: there’s some up there already | 16:04 |
| stickupkid | hml: fair | 16:04 |
| stickupkid | hml: i'll see how that works in a minute | 16:04 |
| rick_h_ | externalreality: are you sure on the do-release-upgrade? Did that get pushed up? I don't see it in the diff and the second round of QA still shows the space. | 18:37 |
| veebers | Morning all o/ | 20:44 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!