[01:28] thumper: https://bugs.launchpad.net/juju/+bug/1765571 [01:28] Bug #1765571: lxd container fails to launch on bionic host: No associated target operation [01:29] thumper: im going to EOD but ill leave those envs up so if you want me to try anything/get more info let me know [01:33] admcleod_: ack [01:33] babbageclunk: ping [01:34] thumper pingback - good timing! [01:34] in 1:1? [01:34] ack [02:52] wallyworld: can you reply to Junien on https://bugs.launchpad.net/bugs/1762979 ? [02:52] Bug #1762979: juju resolve --no-retry behaviour is inverted [02:52] looking [03:30] babbageclunk (thumper?) I found go-guru-hl-identifier-mode, I find it useful you may too (highlights identifier under point). [03:30] veebers: might try it, thanks! [03:32] I enabled with "(add-hook 'go-mode-hook #'go-guru-hl-identifier-mode)", I tweaked the face a little too '(go-guru-hl-identifier-face ((t (:inherit highlight :slant italic)))) [03:47] I like it! It feels a bit inconsistent though - it highlights instaces of time (in time.Duration) but not Duration, and not a different package. Maybe it's just if that name (time) happens to have been used as a local somewhere in the file? [03:48] thumper: :-) pretty please? https://github.com/juju/juju/pull/8634 [03:48] * thumper looks [03:49] need for CI etc [03:49] wallyworld: got a sec to chat about this? [03:49] sure [03:49] 1:! [04:29] hi wallyworld: can I have ur 5mins on hangout? [04:29] kelvinliu: just talking to chris, give me [04:29] 5 [04:29] unless you want to join standup hangout [04:30] wallyworld: sure, I will just join soon. thx [04:32] kelvinliu: i *need* coffee, let me go make one and i'll ping you in 5 [04:33] wallyworld: yup, no problem, thx [04:38] thumper: retrying that benchmarking on aws machines I can get between 340 - 400 updates per second. Weirdly, tweaking that parameter I was talking about didn't seem to make much difference. [04:39] hmm... ok [04:42] kelvinliu: free now, see you in hangout? [04:42] wallyworld: yup [05:09] * thumper sighs [05:09] looks like gofmt is sad on develop [05:09] how? [05:10] did wallyworlds change land for make check? [05:10] type DBusAPIFactory = func() (DBusAPI, error) [05:11] there is an = there [05:11] service/systemd/service.go [05:11] go test passes in there [05:12] but the prepush hook fails [05:13] wallyworld: a quick one https://github.com/wallyworld/caas/pull/2 thx [05:13] https://github.com/juju/juju/commit/4185a33acc39527276de908c2492f6286d87f0d1 [05:13] funny, but Joe's change was just renaming [05:13] kelvinliu: thks [05:14] the type alias is from before [05:14] perhaps this is a new check? [05:14] thumper: which go version are you using? [05:14] 1.10.1 [05:14] I see the prechecks are happening (this run fron ~17 hours ago: http://ci.jujucharms.com/job/github-check-merge-juju/963/console) [05:15] kelvinliu: it includes your bundle changes, they will need to be backed out [05:16] kelvinliu: also build dirs like charms/gitlab/deps/interface/pgsql [05:16] need to not be pushed [05:17] wallyworld: use INTERFACE_PATH=/path-to-ur-local-interfaces-dir use local interface [05:17] kelvinliu: thanks, yeah, i already had that set as it turns out [05:19] wallyworld: ah, ok, I will gitignore deps [05:19] wallyworld: is the deps required or not [05:19] no [05:20] can be removed from PR [05:20] is generated on build AFAIK [05:21] kelvinliu: looks good with a few comments to fix [05:22] wallyworld: https://github.com/juju/juju/pull/8635 [05:30] wallyworld: made changes for the comments. thx [05:31] kelvinliu: where does ##MODEL## come from? [05:31] i've not seen that before [05:31] wallyworld: in the $profile [05:31] the bash profile? [05:32] ih wait [05:32] i see [05:33] wallyworld: yes [05:33] thanks, haven't noticed that before [05:33] wallyworld: np :) [05:34] kelvinliu: looks good, i'll merge when the mysql layer change gets merged so as not to break existing users [05:35] wallyworld: yeah, thx [05:40] wallyworld: hi [05:40] hey [05:40] wallyworld: so afaik, "juju resolved" has been, by default, retrying _by default_ since 2.x (1.x default was to not retry) [05:41] wallyworld: and you're saying in the bug that most 2.x versions didn't retry if one simply ran "juju resolved" ? [05:41] yeah, in 2.x the behaviour was meant to be swapped [05:41] 2.x requires --no-retry to not retry [05:41] 2.x retries by default [05:41] or is supposed to [05:42] but i was testing an found it didn't [05:42] wallyworld: if you are happy with the proxy rework branch, just add the merge comment, otherwise I'll address anything monday morning [05:42] laters peeps [05:42] thumper: looking [05:42] axino: and i looked at the code and saw a bug [05:44] wallyworld: I'm suprised no one saw it in 1+ years [05:45] axino: me too. apparently tim was talking to xav from bootstack and when the bug was mentioned, xav said somrthing like "oh, i was wondering why it never seemed to work properly" [05:45] heh [05:45] better late than never for a fix i guess [05:48] oh yeah definitely [05:48] wallyworld: and this is an agent thing right ? not a client thing ? [05:48] yeah, if i recall correctly [05:49] CLI was modified but agent wasn't [05:49] modified to the reversed behaviour [06:30] kelvinliu: see this error? https://pastebin.ubuntu.com/p/MF3wRbZWFp/ [06:30] the mysql charm relation joined isn't happy [06:31] wallyworld: ah, sorry. I changed the code inside the container by exec into it, but forgot change it into the repo. [06:32] ok, np, just push the change :-) [06:33] wallyworld: done. [06:33] thanks, looking [06:34] wallyworld: np [06:39] kelvinliu: it seems then that he doc string is wrong in the mysql layer [06:39] they leave off the items() also [06:39] wallyworld: yes, I was about to add comment to the PR [06:40] great! [06:40] just waiting for gitlab to start now so i can see the result [06:45] kelvinliu: there is indeed a problem - gitlab can't talk to mysql, gets auth error it seems [06:45] wallyworld: just found one issue on operators. so it seems operator was deployed by k8s pod api, so no `deployment` resources created. So k8s will not be able to manage it for recovering for example. [06:46] you sure? i get deployments [06:46] $ kubectl -n caas get deployments [06:46] NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE [06:46] juju-gitlab 1 1 1 1 6m [06:46] juju-mysql 1 1 1 1 8m [06:47] that's the deployments for app pod [06:47] i mean depployment for operator pod [06:47] oh right yes, deployment for operator pod not done yet [06:47] there's a todo in the code [06:48] ah, ic. I was just deleted gitlab operator pod, so it's gone [06:48] kelvinliu: so if mysql password is being correctly passed to gitlab docker image, but gitlab can't connect, that implies an issue in provides.py on the mysql side i think [06:49] you'll have to remove-application from juju and deploy again [06:49] yup. [06:50] kelvinliu: actually, it looks like an issue with the gitlab image https://pastebin.ubuntu.com/p/2yxHnJffqT/ [06:51] we use latest, i might try an older one [06:51] wallyworld: yes, i got the same error. [06:54] wallyworld: but latest image was pushed 10 days ago === frankban|afk is now known as frankban [07:02] kelvinliu: i tried an older version 10.6.4, same error :-( [07:13] wallyworld: i guess it's related with the db access [07:14] it is but we need to find out why [07:14] will have to try with theold mysql layer [07:16] kelvinliu: solved! [07:16] it was new mysql version 8 [07:16] this works [07:16] juju deploy ./mysql --config mysql_image=mysql/mysql-server:5.7 [07:17] i will push a change to the charm to use that tag [07:18] wallyworld: ah mysql had a recent push. [07:18] yeah [07:20] wallyworld: would be better to lock the version tag [07:20] kelvinliu: that's what i'm going to do - change the tag in the charm config [07:21] config.yaml has the default image path [07:21] in my example above, i overrode the default when deploying [07:30] wallyworld: ah, ic. the charm actually should be latest, then overwrite it from cmd when we deploy. probably we can just add the current compatible versions to the README? [07:32] kelvinliu: wallyworld would not have seen ur message, he logged out just before it. [07:33] anastasiamac: yeah, just saw that. thx for remind. haha [07:34] nws :) [07:34] m guessing u r not using tab to autocomplete the nick :D [07:40] ah, how did u know? i just ctl+c ^-^ [08:04] kelvinli_: i've pushed changes to pin the image version. you could rebase your branch [08:08] kelvinli_: left some more comments in the PR [08:13] wallyworld, looking the comments now [08:26] kelvinli_: thanks for changes! perhaps a comment that the -o directory can be anywhere convenient; doesn't have to necessarily be ~/.local/shared/charms [08:54] wallyworld, yeah, agreed. [08:54] \o/ [09:06] Have a good weekend :-) [12:35] Good morning === frankban is now known as frankban|afk [18:45] Bug #1729930 changed: juju.state.leadership manager.go:72 stopping leadership manager with error: state changing too quickly; try again soon

[20:18] balloons: regarding the prehook checks not working on first push of new branch: https://paste.ubuntu.com/p/jV84x29pZm/ [20:19] balloons: perhaps because master is now not used? guessing in the dark [20:19] Ah, yea we should remove master and staging [20:20] hml, did you try a quick fix of looking in develop? [20:20] balloons: sure [20:21] I would rather we must added the meta linter instead of that old script [20:25] balloons: i did a simple change of master to develop and still fails… something else is wrong.. not a git master [20:26] Gotcha. Sorry it's a pain