timClicks_ | babbageclunk: hey sorry to bother you again, would you mind helping me out with an interface/type coercion issue I've encountered | 02:12 |
---|---|---|
babbageclunk | timClicks_: sure sure! | 02:12 |
timClicks_ | I'm 90% sure that I've identified the cause of the import issues | 02:12 |
babbageclunk | oh awesome | 02:12 |
thumper | timClicks_, babbageclunk: what's the tl;dr on vsan? | 02:14 |
timClicks_ | thumper: tell you in a few minutes | 02:14 |
thumper | ack | 02:15 |
timClicks_ | thumper: we've identified what's causing the cloud-init the issue; specifying user-data is available into a VirtualApp (container for 1 or more VMs) rather than a VM | 02:37 |
timClicks_ | annoyingly, the call to ImportVApp (which we | 02:38 |
timClicks_ | we have been using) returns a VM if the OVF only defines a VM | 02:38 |
timClicks_ | it looks like we need to customise the OVF XML file that we embed | 02:39 |
timClicks_ | babbageclunk: it looks like we need to wrap the <VirtualSystem> tag within a <VirtualSystemCollection> | 02:47 |
babbageclunk | timClicks_: yeah that makes sense | 02:47 |
timClicks_ | babbageclunk: tweaking the ovf is creating a vapp | 03:07 |
timClicks_ | *babbageclunk: tweaking the ovf is working, currently creating a vapp | 03:07 |
babbageclunk | timClicks_: I got distracted by the following rabbit-hole: VirtualMachineConfigSpec.VAppConfig.Property (which looks like it would let you edit the fields on that form in the clone VM wizard) | 03:30 |
timClicks_ | babbageclunk: that looks promising | 03:30 |
timClicks_ | using a vapp directly doesn't seem like a good option | 03:31 |
babbageclunk | Having dug around a bit I think extraConfig is still actually what we want though - trying that now | 03:31 |
babbageclunk | timClicks_: bums - yeah, I was worried about that | 03:31 |
timClicks_ | babbageclunk: we have a method to modify extra config in our vsphereclient code | 03:32 |
babbageclunk | bugger, looks like I'm still uploading a vmdk | 03:32 |
babbageclunk | looking for that | 03:32 |
timClicks_ | babbageclunk: it looks like you've been able to power on a machine? | 03:49 |
babbageclunk | yeah, but no dice - it still lists the same values in its vapp options | 03:50 |
babbageclunk | ok, I'm going to have a go at changing the VAppConfig stuff. It's all a bit black magic though | 03:55 |
babbageclunk | timClicks_: trying using govc to extract info about the vm so I can feed that back in | 04:08 |
wallyworld | kelvinliu_: there's a k8s bug :-( https://github.com/juju/juju/pull/10515 | 05:17 |
kelvinliu_ | wallyworld: looking | 05:18 |
kelvinliu_ | wallyworld: lgtm, thanks! | 05:25 |
wallyworld | ty | 05:25 |
wallyworld | stupid typo | 05:25 |
kelvinliu_ | [Facepalm] | 05:28 |
stickupkid | rick_h, achilleasa here is the output of a bash cmr bundle example https://paste.ubuntu.com/p/k2S8gTbBJk/ | 14:03 |
stickupkid | achilleasa, it works really nicely | 14:03 |
stickupkid | the great thing is, that test from e2e ran in 2m28s which beats the hell out of the python one | 14:08 |
gizmo693 | Hi Guys. Newbie question. I am working the the docs on https://jaas.ai/docs/getting-started-with-charm-development and when I get to the charm build command it is failing | 14:20 |
gizmo693 | According to magicaltrout there is a missing series in the metadata.yml ? | 14:20 |
gizmo693 | Any pointers/advice? | 14:20 |
rick_h | gizmo693: the metadata.yaml needs a series key to say what ubuntu series the charm works on. | 14:21 |
gizmo693 | I have a trout that is magical in my office so will be able to ask what that means | 14:22 |
stickupkid | gizmo693, example https://api.jujucharms.com/charmstore/v5/~jameinel/ubuntu-lite-7/archive/metadata.yaml | 14:22 |
gizmo693 | From a newbies perspective how would I (anyone) know that | 14:23 |
stickupkid | gizmo693, good shout tbh | 14:23 |
rick_h | gizmo693: if you're using a starting template I think it defaults to the latest stable series for you | 14:23 |
rick_h | gizmo693: but the question is what os/version does the software your charm works against work on? | 14:23 |
gizmo693 | I'm just trying to work through the getting started guide as opposed to actually building a charm of my own | 14:25 |
stickupkid | achilleasa, export bundle test https://github.com/SimonRichardson/juju/blob/tests-suite/tests/suites/cmr_bundles/export_overlay.sh#L1-L36 | 14:25 |
gizmo693 | magicaltrout said is it charm build or charm create I should be using? | 14:26 |
achilleasa | stickupkid: looks nice! One neat trick that I like to use is to pipe the output to something like "sed -e 's/^/ |/g'" so you can get it indented | 14:27 |
rick_h | gizmo693: ok, if you've got comments/feedback/edits please note the link at the bottom which goes to the discourse post for the page you're on. https://discourse.jujucharms.com/t/getting-started-with-charm-development/1118 | 14:27 |
rick_h | gizmo693: charm create will let you start with a known template | 14:27 |
rick_h | gizmo693: charm build is the command to build your charm (compile it really) | 14:27 |
stickupkid | achilleasa, i was thinking about that, but then i'm wondering if we should do it everywhere? | 14:28 |
achilleasa | stickupkid: possibly; or create library functions that you include in your script (ala init.d's functions) | 14:29 |
stickupkid | achilleasa, yeah, that's easy to do, anything in side this folder gets included across all tests https://github.com/SimonRichardson/juju/tree/tests-suite/tests/includes | 14:30 |
achilleasa | stickupkid: any plans for handling timeouts? (for juju deploy) | 14:34 |
stickupkid | achilleasa, we could easily use timeout | 14:35 |
achilleasa | stickupkid: +1 | 14:35 |
stickupkid | achilleasa, i.e. `timeout -k 10m juju bootstrap lxd test` | 14:35 |
rick_h | achilleasa: time to catch up? | 14:36 |
achilleasa | rick_h: sure | 14:36 |
rick_h | achilleasa: k, meet you in daily | 14:37 |
rick_h | stickupkid: "no description provided" boooooo :P | 15:05 |
stickupkid | rick_h, for? | 15:05 |
rick_h | stickupkid: your PR | 15:05 |
stickupkid | rick_h, ah, i'll fill that in now | 15:05 |
stickupkid | rick_h, done | 15:12 |
rick_h | stickupkid: k, review done | 15:13 |
rick_h | stickupkid: anything else you needed in particular for today? | 15:13 |
stickupkid | rick_h, sometime to quickly have a look at the bash stuff | 15:14 |
rick_h | stickupkid: k, that's going to go up as a PR of sort right? | 15:15 |
stickupkid | rick_h, yeah | 15:15 |
rick_h | stickupkid: ok, let me know when that's up and will peek at it | 15:15 |
stickupkid | rick_h, i'll make that PR then, let me close out the pylib juju | 15:15 |
stickupkid | rick_h, sure | 15:15 |
stickupkid | rick_h, looks like https://github.com/juju/python-libjuju/pull/336 is good to me | 15:19 |
rick_h | stickupkid: cool, aisrael going to hit merge on that then | 15:21 |
aisrael | rick_h: stickupkid: ta, thanks! | 15:22 |
achilleasa | rick_h: the async charm stuff works with macaroons as well :-) | 15:58 |
stickupkid | this annoys me https://paste.ubuntu.com/p/Yxq6sjpg93/ - juju returns something on stderr in none json format when I ask it to do so | 16:20 |
stickupkid | doing this makes me sad "juju status --format=json 2> /dev/null" | 16:21 |
rick_h | achilleasa: woot | 16:31 |
rick_h | stickupkid: seems less than ideal | 16:31 |
stickupkid | rick_h, esp. considering we're now hiding error messages :( | 16:32 |
stickupkid | rick_h, at some point in the future it would be nice to add better ux/design to the CLI - it's a bit rough and showing it's age esp. when comparing snap, kubectl etc | 16:35 |
rick_h | stickupkid: yea, I think that's goals for the 3.x flow | 16:35 |
stickupkid | rick_h, let's go mental and add graphs https://github.com/gizak/termui | 16:35 |
stickupkid | :wink: | 16:35 |
rick_h | lmao | 16:35 |
thumper | morning | 20:31 |
rick_h | morning thumper | 20:31 |
wallyworld | babbageclunk: timClicks: what's the goss on vsan? | 21:27 |
babbageclunk | wallyworld: some progress - but have hit another snag, not sure what the issue is | 21:36 |
wallyworld | ok, i'll try and organise access to a vsphere technical person whom we can enage with | 21:37 |
babbageclunk | wallyworld: that would be awesome | 21:37 |
babbageclunk | wallyworld: things are looking pretty good | 21:57 |
babbageclunk | wallyworld: charmed-k8s seems to be deploying happily | 22:06 |
babbageclunk | (all machine agents are up, waiting for charm installs) | 22:06 |
timClicks | wallyworld, thumper: looks like babbageclunk's nailed it | 22:32 |
wallyworld | oh, goody. i'll catch up and get the details soon | 22:32 |
babbageclunk | wallyworld: it looks like our vsphere machines can't install snaps though | 22:32 |
babbageclunk | might need a network change | 22:33 |
wallyworld | can you create an RT? | 22:33 |
timClicks | on it | 22:33 |
wallyworld | ty | 22:33 |
timClicks | just confirming that we haven't somehow introduced it via the patch by reproducing on 2.6.7 | 22:33 |
thumper | what snaps are we installing? | 22:34 |
wallyworld | cdk uses snaps | 22:35 |
thumper | ah, right | 22:35 |
timClicks | as a side issue, using juju 2.6.7 to deploy multiple new units feels comically slow compared to the new patch | 23:18 |
timClicks | babbageclunk: the cdk install issue isn't due to us | 23:36 |
babbageclunk | timClicks: woot | 23:37 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!