wallyworldhpidcock: for whenever, here's that PR with the action data mutex added https://github.com/juju/juju/pull/11416. protects acess from the hook commands and the uniter itself during and after running an action00:59
hpidcockwallyworld: approved01:00
thumperdo we have any focal deployment tests in CI yet?01:04
wallyworldthumper: i know we added focal support to all the test charms back in late feb, not sure about what else01:33
hpidcockwallyworld: we don't use any cgo specifics from the github.com/coreos/go-systemd package that I can see02:03
hpidcockI take that back02:05
wallyworldno backsies02:06
hpidcockgithub.com/juju/juju/service/systemd uses github.com/coreos/go-systemd/util.IsRunningSystemd02:06
hpidcockoh phew02:07
hpidcockthat uses a file, ALL good!02:07
hpidcock10/10 no problems02:07
hpidcockthe snap is 125Mi02:13
hpidcockso I think this is a winner02:13
hpidcockbootstraps to lxd fine02:13
wallyworldgreat, you checked the jujud binary on controller machine just to be sure?02:14
hpidcockhave now :)02:21
hpidcockwallyworld: https://github.com/juju/juju/pull/1141702:29
hpidcockor anyone else02:29
wallyworldhpidcock: we can get rid of the x-dep plugin when we get to go mod :-)02:30
hpidcockwallyworld: yeah I think its probably best we will have our fork and see if we can get some features merged back into the official plugin02:31
hpidcockI've tried to keep the parameters the same where I can, but there are features they don't expose we might want to update in the upstream plugin02:32
wallyworldtrue, although i'm not sure people care about that plugin anymore02:32
hpidcockno I mean the go plugin not the dep one02:32
wallyworldah right02:33
wallyworldkelvinliu: small fixes for beta1 https://github.com/juju/juju/pull/1141803:11
kelvinliulgtm ty03:17
wallyworldhpidcock: those jenkins qa changes merged,i guess you are going to run jjb?03:17
hpidcockI've been iterating, just wanted to get my fixes landed03:17
hpidcockwas banging my head for more than the minutes it should have taken03:17
wallyworldno worries, just duble checking to ensure nothing slips by03:18
stickupkidmanadart, you got a sec?09:12
manadartstickupkid: Yep.09:14
stickupkidin daily09:16
manadartachilleasa: Got a sec to look at https://github.com/juju/juju/pull/11421 ?09:35
achilleasamanadart: looking09:36
achilleasamanadart: approved09:53
manadartachilleasa: Thanks.09:53
achilleasamanadart: can you take a look at https://github.com/juju/juju/pull/11422?10:20
manadartachilleasa: See how I go. I have to hit the road soon, but I'll look while my patch is landing.10:21
achilleasamanadart: not in a hurry; got one more following up. If busy I can ask Heather to take a look when she gets online10:22
manadartachilleasa: Approved it.10:48
rick_h_stickupkid:  https://bugs.launchpad.net/juju/+bug/1730747 might be of interest to that reload-spaces work. I think we should be able to knock a few off the list12:33
mupBug #1730747: incomplete support for spaces in juju and maas <maas-provider> <spaces> <sts> <juju:Triaged> <https://launchpad.net/bugs/1730747>12:33
achilleasaanyone up for a small review? https://github.com/juju/juju/pull/1142312:39
stickupkidrick_h_, about 5-6 bugs in one :)12:49
rick_h_stickupkid:  yea, exactly. Let's see if we can knock a couple out and maybe do individuals from there with tags for spaces and such for the future?12:50
stickupkidachilleasa, can look into testing non-modified part.13:21
achilleasastickupkid: got a question about your PR comment13:36
stickupkidachilleasa, the former. I want see that we don't call SetState13:37
achilleasastickupkid: but that's essentially testing the Modified method, right?13:38
stickupkidachilleasa, well,13:38
achilleasano objection to adding another test; just trying to understand what value we get out of it13:38
stickupkidachilleasa, well I find it ensures that we're changing functionality based on a state13:39
stickupkidachilleasa, generally test that13:39
stickupkidachilleasa, as apposed to testing every error case13:40
achilleasastickupkid: pushed a commit with the extra test; can you take a look?14:15
stickupkidachilleasa, already ticked14:22
stickupkidrick_h_, achilleasa can you delete the alpha space?14:32
* stickupkid hopes the answer is no14:32
achilleasastickupkid: my assumption is that it is read-only14:32
achilleasaI mean, the instancepoller puts virtual IPs into alpha; if you deleted that were would they go if everything must belong to a space?14:33
achilleasa(because VIPs don't match any of the CIDRs)14:33
stickupkidachilleasa, yeah chaos indeed14:33
rick_h_stickupkid:  nope14:56
stickupkidwicked, this is great news14:56
rick_h_stickupkid:  alpha space is the default, always there, home for things not assigned14:56
thumperbabbageclunk: are you still after a review for PR 11426?23:16
babbageclunkthumper: yes please23:16
thumperok, I'll look now23:16
thumperbabbageclunk: code generally looks good, just one request around behaviour when things don't go well23:29
babbageclunkok thanks23:29
babbageclunkthumper: I don't know about that - *in theory* there could be another change to the uniter state file later that would need a different upgrade step to migrate - if this one goes nuclear and trashes the file for changes it doesn't know about that might be a bad thing?23:35
babbageclunkthumper: definitely a theoretical problem though23:36
thumperperhaps instead of using the validity function, perhaps just read the yaml, look for the remote-unit key, and if it is there add the remote-application one?23:37
thumperthat way we avoid the validity issue23:37
thumperI think you are right on that aspect23:37
babbageclunkok, I'll tweak it to do that. Actually what's there now won't work in the scenario I'm describing, the Write call will fail if the state's invalid after a change.23:38
babbageclunkSo it needs to be done anyway23:38
babbageclunkI'll leave the validity check on the way in, but make sure we can still write the change even if the resulting state isn't (yet) valid.23:40
thumperyeah, I guess the initial validity check is fine, because if it is valid, there is nothing to do23:41
thumperI was just wondering how best to deal with the issues...23:41
thumperbut I think you are right, there is potential that a later release would cause an invalid file in a different way23:42
thumperand that this upgrade step isn't there to fix that23:42
thumperbut it would be called first if they are upgrading through multiple versions23:42
babbageclunkMaybe a post-upgrade-steps sanity check? Just reads the state, and if it's bad then throws it away?23:42
thumperso I agree with you now that deleting it isn't the right thing to do23:42
thumperno... I don't think so23:42
thumperconsider being invalid for two reasons?23:43
thumper2.7 added remote-application23:43
babbageclunkI mean, after all of the steps are run23:43
thumperbut say 2.8.1 adds another key that is needed23:43
thumperso I don't think we should throw away the change just because the result isn't yet valid23:43
thumperbecause we can't guarantee that a future step wouldn't fix it23:43
babbageclunkright, but there's a point where all the available upgrade steps have run, if the state's invalid at that point then maybe we need to throw it away23:44
thumperbut instead of reading and writing using a structure, I'd recommend just using a map[string]interface23:44
thumperah... I see what I think you are getting at23:44
thumperwe don't have anything like that yet...23:45
babbageclunkyeah, thinking about it I should be reading into a map, just in case the current defn of State's changed so much that I can't load the file as is.23:46
babbageclunkwill change that too23:46
timClicksis anyone with experience with reactive charms able to provide some advice here? https://discourse.juju.is/t/workload-stuck-in-maintenance-status/289023:46

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!