| [01:34] <pmatulis> i remember asking about that a year ago. iirc, they are provider-dependent | |
| [01:57] <timClicks> pmatulis: ty | |
| [01:57] <timClicks> that's the conclusion I have come to | |
| [02:28] <wallyworld> thumper: https://github.com/juju/juju/pull/11703 | |
| [02:29] * thumper headdesks at 2.8 branch needing apiserver to test state package | |
| [02:29] * thumper looks at pr | |
| [02:29] <thumper> igmores | |
| [02:30] <thumper> 15 files? | |
| [02:30] <wallyworld> small changes | |
| [02:32] * thumper nods | |
| [02:35] <thumper> ah FFS | |
| [02:35] * thumper wonders about 2.7 all watcher bits | |
| [02:36] <thumper> wallyworld: this isn't going to work | |
| [02:36] <thumper> did you want to jump in a HO to discuss | |
| [02:36] <wallyworld> sure | |
| [02:36] <thumper> 1+1 | |
| [03:31] <wallyworld> thumper: tweaked the cache lookup, just need to test with libjuju | |
| [03:32] <wallyworld> and need to fix one unit test | |
| [04:12] <thumper> wallyworld: is it ready for a rereview? | |
| [04:13] <wallyworld> thumper: HO? | |
| [04:27] <thumper> sure | |
| [04:48] <thumper> hpidcock: did your new multichecker land in 2.7? | |
| [04:48] <thumper> hpidcock: I have a perfect usecase for it | |
| [04:48] <hpidcock> thumper: I don't think so | |
| [04:49] <thumper> hpidcock: boo | |
| [04:49] <hpidcock> Feel free to update the juju/testing dep | |
| [04:50] <thumper> hpidcock: hmm... | |
| [04:50] <thumper> hpidcock: may well do | |
| [04:50] <thumper> hangon | |
| [04:50] <thumper> I'm in 2.8 | |
| [04:50] <thumper> yay | |
| [04:50] <thumper> hpidcock: what is it called again | |
| [04:51] <hpidcock> jc.NewMultiChecker() | |
| [04:51] <hpidcock> there is an example of its use somewhere in 2.8 | |
| [04:52] <thumper> found it | |
| [04:52] <thumper> now I need to work out how to get it to ignore the right field | |
| [04:53] <thumper> hpidcock: got 2 minutes to help out? | |
| [04:55] <thumper> I'm not sure how to map this thing through | |
| [04:55] <hpidcock> sure | |
| [04:55] <hpidcock> standup? | |
| [04:56] <hpidcock> thumper:^ | |
| [04:57] <thumper> hpidcock: yeah | |
| [05:03] <wallyworld> thumper: you good with the PR then? i wan tto land it so i can forward port | |
| [05:28] <thumper> wallyworld: sorry, will do another quick pass over | |
| [05:28] <wallyworld> all good | |
| [05:32] <thumper> and you push while I'm looking? | |
| [05:32] <wallyworld> thumper: sorry, there was a go fmt thing | |
| [05:33] <wallyworld> an extra ' ' | |
| [05:34] <thumper> approved with a questino | |
| [05:34] * thumper is done and heading off now | |
| [05:34] <thumper> leaving with 15 failing tests in state package | |
| [05:34] <thumper> get to them next week | |
| [05:34] <thumper> later peeps | |
| [09:05] <achilleasa> manadart: to fix the lingering port issue with ovs we could override the port name via the 'veth.pair' option to a known (based on the container id or similar) value so we can remove it when we destroy the container | |
| [09:05] <manadart> achilleasa: Yep. | |
| [09:46] <stickupkid> yay, managed to replicate the LXD bug, finally | |
| [10:20] <achilleasa> now, we have deterministic iface naming for lxd veths ;-) https://pastebin.canonical.com/p/ysbqGHWJVV/ | |
| [10:21] <achilleasa> manadart: stickupkid comments/concerns about format? ^^ (it's "machID-ifaceIdx" where machID has / converted to _ to distinguish from last dash) | |
| [10:22] <stickupkid> achilleasa, i'm never sure if you even need _, but otherwise looks ok to me | |
| [10:23] <stickupkid> achilleasa, i.e. 0lxd2-0 | |
| [10:23] <stickupkid> achilleasa, lxd is the separator | |
| [10:23] <stickupkid> achilleasa, like doing `1_,_2_,_3` for csv ;-) | |
| [10:24] <manadart> achilleasa: Agree with stickupkid. This is cool BTW. | |
| [10:25] <manadart> achilleasa: This will apply to all LXD veths created for Juju right? not just connected to OVS? | |
| [10:29] <achilleasa> stickupkid: makes sense; I will just trim it off. manadart yes; I wanted to put a juju prefix but there is a 16 char limit so I tried to keep it shorter | |
| [10:31] <manadart> achilleasa: I love that "ip a" will now give a less opaque picture of what's going on. | |
| [10:31] <manadart> achilleasa: And in the state link-layer data for that matter. | |
| [15:20] <hml> stickupkid: pr review changes made to 11701. i’m finalizing k8s testing now | |
| [15:41] <stickupkid> hml, been looking | |
| [15:41] <hml> stickupkid: testing my k8s myself | |
| [15:47] <stickupkid> hml, approved again | |
| [15:48] <hml> stickupkid: cool, ty | |
| [15:49] <hml> stickupkid: ffs: https://pastebin.canonical.com/p/J3Tx9pF6Wc/. will have to fix this | |
| [15:52] <hml> stickupkid: and k8s is now failing to install *sigh | |
| [15:53] <stickupkid> hml, weird | |
| [15:54] <hml> stickupkid actually no… the LXDProfileRequire doesn’t check machine for KVM or manual etc…. only if the charm has a profile in it | |
| [15:55] <hml> *facepalm* | |
| [15:55] <hml> not sure on the k8s change yet | |
| [16:23] <stickupkid> hml, if you've got 5 minutes before my EOD around https://github.com/juju/juju/pull/11700 | |
| [16:24] <hml> stickupkid: had a question for you on there… forgot to click send. ho? | |
| === mpontillo_ is now known as mpontillo | |