[00:00] <kelvinliu> ty hpidcock
[00:00] <kelvinliu> wallyworld: yes, I will have a pr for 2.7
[00:01] <wallyworld> no need for a 2.7 fix, they claimed it was working in 2.7
[00:01] <wallyworld> so i was curious if we had introduced a regression in 2.8
[00:06] <wallyworld> kelvinliu: i think we're ok to $$mergre$$ into 2.8?
[00:08] <kelvinliu> wallyworld: we have this for legacy podspec in 2.7
[00:08] <kelvinliu> and podspecV2 in 2.8
[00:09] <wallyworld> so v3 is ok?
[00:09] <kelvinliu> latest/current podspec has no issue
[00:10] <kelvinliu> coz only previous version can have ProviderPod == nil
[00:10] <wallyworld> let's land in 2.8 so we can get the release candidate built. we're not doing any more 2.7
[00:10] <kelvinliu> yep
[00:44] <wallyworld> kelvinliu: here the port of the 2.8 fix https://github.com/juju/juju/pull/12143
[00:47] <kelvinliu> wallyworld: lgtm thx for doing this port pr
[00:47] <wallyworld> np
[01:06] <thumper> o/
[01:06] <thumper> how's things going?
[01:07] <wallyworld> getting 2.8.6 candidate packaged today, 2.9 beta as well, so busy
[01:08] <thumper> how's the new k8s stuff going?
[01:08] <wallyworld> having "fun with leadership/locks" on heavily loaded microk8s deployments
[01:08] <wallyworld> 2.9 beta will allow a POC/MVP charm with v2 metadata to be tried
[01:10] <thumper> nice
[01:10] <thumper> how does it feel to use?
[01:11] <wallyworld> hard to say because it's not much more than a simple pod with a stateful set. so all the other things people use not there yet
[01:12] <wallyworld> v2 metadata much better though (not just for k8s), it ditches series in favour of os/version/arch
[02:47] <wallyworld> kelvinliu: to help debug the k8s dex-auth issue, i've added extra logging https://github.com/juju/juju/pull/12144
[02:47] <wallyworld> we'll also want to run the kubeflow CI tests with TRACE turned on for
[02:47] <wallyworld> juju.worker.raft
[02:47] <wallyworld> juju.worker.raft.raftforwarder
[02:47] <wallyworld> juju.worker.lease.raft
[02:48] <kelvinliu> looking
[02:55] <kelvinliu> wallyworld: lgtm, will just set the root to trace
[02:55] <wallyworld> that will be a lot of trace, might load it up even more
[02:57] <kelvinliu> we only have those logs uploaded if the tests failed
[03:01] <kelvinliu> so it might be ok, u know we can't change this whenever we want see something different,
[03:08] <wallyworld> kelvinliu: yeah, we can't change, but my thought was trace for everything would make things worse, as opposed to just trace for those selected packages. but we can try with trace for root and see
[03:12] <kelvinliu> yeah the logger sender would be more busy
[03:55] <wallyworld> kelvinliu: ah, i landed that logging PR to develop not 2.8. will have to do another PR
[03:56] <kelvinliu> ok
[03:57] <kelvinliu> wallyworld: u wanna change the log level for both controller and model, right?
[03:57] <wallyworld> yeah, lets
[03:58] <wallyworld> kelvinliu: here's the backport https://github.com/juju/juju/pull/12145
[04:00] <kelvinliu> lgtm
[04:01] <kelvinliu> just set the logging level to trace for ken's pr
[04:01] <wallyworld> great ty, i'll get the 2.8 edge snap built as soon as the PR lands
[04:02] <wallyworld> hopefully we can then trigger a run manually
[08:04] <manadart> achilleasa: Can you take another look at https://github.com/juju/juju/pull/12118 ? I reverted the change to return NoSupported, because there are so many test suites that use the method. I'll use an alternative and clean it all up in a later patch.
[08:26] <achilleasa> manadart: still got a failing client test
=== mup_ is now known as mup
[10:04] <achilleasa> manadart: or stickupkid can either of you take a quick look at a small typo fix? https://github.com/juju/juju/pull/12147
[10:26] <stickupkid> manadart, I've just disabled the test for now https://github.com/juju/juju/pull/12148
[10:27] <manadart> stickupkid: Ack.
[10:32] <stickupkid> manadart, https://github.com/juju/juju/pull/12149
[13:09] <elox> Hey, I'm struck by a bug I would like some help in understanding how to resolve. I get an error from pip when trying to install PyYAML which seems to come from the reactive framework. It blows up my installation of nextcloud. Here is a screenshot of the error. https://pasteboard.co/JvUDbhx.png
[13:38] <Chipaca> elox: looks like the error is from trying to _uninstall_ pyyaml
[13:38] <Chipaca> anyway, probably not helpful :-/
[14:04] <manadart> Needs review: https://github.com/juju/juju/pull/12151