[00:00] <hpidcock> kelvinliu: it wasn't added in the rc branch
[00:01] <hpidcock> only the 2.8 branch
[00:01] <kelvinliu> ah, ic
[00:08] <hpidcock> wallyworld: did you want me to quickly look at both issues and land in 2.8?
[00:08] <wallyworld> hpidcock: if you had time that would be grand
[00:08] <hpidcock> np
[02:01] <pmatulis> does the '--map-machines' option need to be used when deploying with both a bundle and an overlay? i'm thinking that the bundle will define the existing machines and then the overlay needs the option. or is an existing machine completely "pre-deploy"?
[04:25] <hpidcock> wallyworld: https://github.com/juju/juju/pull/11720
[04:26] <hpidcock> ended up chasing my tail, thought I saw really small binaries of 40mb etc,  for some reason was unable to repro
[04:31] <wallyworld> looking
[04:33] <wallyworld> hpidcock: lgtm ty
[04:33] <wallyworld> i shold merge directly
[04:34] <wallyworld> i'll let the check run pass
[04:46] <wallyworld> hpidcock: merged
[05:09] <timClicks> stupid question sorry... is --device just a k8s thing?
[05:32] <kelvinliu> timClicks: yes
[09:01] <manadart> achilleasa: QA is failing for me on #11716
[09:01] <mup> Bug #11716: psycopg: new changes from Debian require merging <psycopg (Ubuntu):Fix Released by amu> <https://launchpad.net/bugs/11716>
[09:02] <manadart> Stupid mup. The PR, not LP.
[09:02] <manadart> mup--
[09:06] <Chipaca> juju#11716
[09:06]  * Chipaca guesses
[09:06]  * Chipaca guesses poorly
[09:06] <achilleasa> manadart: ho?
[09:07] <manadart> achilleasa: OMW
[09:50] <achilleasa> manadart: pushed the acpi fix to all kvm PRs; taking a look at the failed QA now
[10:11] <stickupkid> CR anyone https://github.com/juju/juju/pull/11715
[10:13] <manadart> stickupkid: I'll do it, then swing back to achilleasa's stuff.
[10:15] <stickupkid> manadart, run this https://github.com/juju/juju/pull/11715#issuecomment-645285581 if you intend to run it all
[10:15] <stickupkid> manadart, life is too short
[10:25] <manadart> stickupkid: Running now.
[10:35] <manadart> stickupkid: Done. You still looking at mine?
[10:35] <stickupkid> manadart, nearly done
[11:00] <achilleasa> manadart: hmmm... so the remove call is skipped because lxd reports the nic type as "broadcast" instead of "bridged"...
[11:02] <manadart> achilleasa: Yeah, I added some Criticalfs here was surprised they weren't hit. So it's bugging out earlier.
[11:02] <achilleasa> also, the lxd from snap on focal seems to remove the entry from the bridge as expected...
[11:02] <achilleasa> yay :D
[11:02] <achilleasa> so I need to tweak some things and re-test on bionic
[11:03] <manadart> achilleasa: I suspect this difference goes hand-in-hand with the profile changes.
[11:04] <manadart> So we get 3.0.3 on Bionic and the latest Snap on Focal.
[11:04] <achilleasa> I will amend my comments accordingly
[11:27] <achilleasa> manadart: updated the PR and verified it works on bionic (I also move the maybeRemove call a bit higher up); can you take another look?
[11:27] <manadart> achilleasa: Yep, saw. Upgrading now.
[11:34] <achilleasa> manadart: if you set juju.core.networking level to DEBUG you should see a log of the call to remove the veth
[11:40] <manadart> achilleasa: Yep, that one's done. Onto the next.
[12:23] <manadart> achilleasa: You try and merge the wrong PR?
[12:24] <achilleasa> manadart: yeah. cancelled the merge once I spotted it... too many open tabs
[12:24] <achilleasa> the right one just landed
[15:32] <hml> stickupkid: i think i know why height and width are not unmarshalling in Media… they can return integer or null
[15:33] <hml> stickupkid: pieced that info out of http://api.snapcraft.io/docs/charms.html
[15:34] <stickupkid> hml, any reason why we read the resp.Body twice?
[15:35] <hml> stickupkid: not that i can think of.
[15:35] <hml> stickupkid: missed that in putting together the spike
[15:35] <stickupkid> hml, as we're not doing auth yet, I'm going to cull most of the do command and bakery client for now
[15:36] <stickupkid> hml, we can add as required (which I'm sure will be soon)
[15:36] <hml> stickupkid: will it be harder to move back in when we do?
[15:36] <stickupkid> hml, nah, just align to an interface
[15:36] <hml> stickupkid: rgr
[15:37] <stickupkid> hml, so the http client and the bakery client both conform to `func (Client) Do(request *http.Request) (http.Response, error)`
[15:38] <stickupkid> hml, so as long as we do that, we can inject what ever we want
[15:38] <hml> stickupkid: ack
[21:26] <cory_fu_> wallyworld: Hey, for CMR, is there a way to get the remote side's model UUID?
[21:34] <cory_fu_> wallyworld: Actually, I should step back.  What I really want to know is when a CMR is created on AWS, how are the SGs modified so that the two units can talk to each other?  Are specific rules added to the unit SGs?
[21:40] <timClicks> does the juju/gnuflags package support modifying a Var once it's been added to the flagset?
[21:42] <timClicks> the juju deploy command includes flags from add-unit directly.. I would like to add " (charm only)" to their help lines when someone execs `juju help deploy`
[21:55] <cory_fu_> wallyworld: Nm, tested and confirmed that it's the specific rules on the unit SG, which makes sense.
[22:41] <timClicks> thumper: I keep wanting to make juju help topics a thing again
[22:41] <timClicks> juju help charm-urls
[22:45] <thumper> timClicks: I'm ok with that...
[22:52] <timClicks> I'm going in circles trying to make 'juju help deploy' shorter. It's getting easier to read, but longer.
[23:48] <tlm> wallyworld: https://github.com/juju/juju/pull/11723
[23:49] <wallyworld> ok
[23:49] <wallyworld> tlm: add bug to PR description, make as in progress...
[23:49] <tlm> ah yep
[23:51] <wallyworld> tlm: just to test end-end, might be work hacking an existing charm and try it, as we've seen once or twice a missing bit in practice, just to be sure etc
[23:51] <wallyworld> it looks ok, but you never know
[23:52] <tlm> any suggestions what what charm to hack etc ?
[23:52] <wallyworld> i use a mariadb one i have unpacked locally. ~juju/mariadb-k8s
[23:53] <wallyworld> cs:~juju/mariadb-k8s
[23:53] <wallyworld> just hack the extra line in the yaml