mbruzek | hello rick_h_ | 14:54 |
---|---|---|
mbruzek | I have a question about your response to https://github.com/CanonicalLtd/jujucharms.com/issues/158 | 14:54 |
mbruzek | please ping me when you have time to talk | 14:55 |
hatch | hey mbruzek did you ever get that GUI issue solved? | 14:55 |
rick_h_ | mbruzek: shoot | 14:55 |
mbruzek | rick_h_: I get that charmworld does not support v4 bundle format yet. But what confuses me is the bundle.yaml found at https://code.launchpad.net/%7Echarmers/charms/bundles/plumgrid-ons/bundle looks to me as a v3 bundle format, because I see the bundle name at the top. | 14:57 |
rick_h_ | mbruzek: ah, but hte filename is wrong for v3, charmworld only looks for bundles.yaml | 14:57 |
mbruzek | If I am mistaken please educate me, but I thought v4 bundle format == no name | 14:57 |
rick_h_ | (with an s) | 14:57 |
rick_h_ | mbruzek: and yes you're right | 14:57 |
rick_h_ | so the bundle is all kinds of confused | 14:57 |
mbruzek | ack | 14:57 |
rick_h_ | mbruzek: honestly I didn't open it, I just looked at the file listing and saw bundle.yaml and replied | 14:58 |
mbruzek | rick_h_: OK. I was just confused. | 14:58 |
rick_h_ | mbruzek: but yes, atm we've got contention with not killing off charmworld/v3 api yet and it's causing this issue in migrating | 14:58 |
rick_h_ | mbruzek: for good reasons, it's not simple and I apologize. Once we kill of v3 it'll be normal | 14:58 |
mbruzek | So your answer is still have both v3 and v4 bundle for now until we can fix the system? | 14:59 |
rick_h_ | mbruzek: you just need enough v3 to make charmworld happy | 14:59 |
rick_h_ | mbruzek: the charmstore will default to using the v4 if it sees it | 14:59 |
rick_h_ | mbruzek: and ignore the v3 | 14:59 |
mbruzek | Ok | 14:59 |
rick_h_ | mbruzek: it could just be a v3 file with one service and nothing else tbh | 14:59 |
mbruzek | In another bundle we have a v3 bundle named bundles_v3.yaml and we have a bundle.yaml | 15:00 |
rick_h_ | mbruzek: did that ingest properly? | 15:00 |
mbruzek | Is that sufficient or do we need to name the v3 bundle, bundles.yaml. | 15:00 |
rick_h_ | charmworld only looks for 'bundles.yaml' | 15:00 |
rick_h_ | so I would not expect anything else to work | 15:00 |
mbruzek | https://jujucharms.com/u/kubernetes/kubernetes-cluster | 15:01 |
mbruzek | it appears to work but if you recommend naming it bundles.yaml that is what we will do | 15:01 |
mbruzek | and by work, I mean I can see it on the web site. | 15:02 |
rick_h_ | mbruzek: did that go through ingestion or get uploaded? | 15:02 |
mbruzek | hatch: no we never resolved that GUI bug, and the customer did not open a bug for it. | 15:02 |
rick_h_ | mbruzek: that's the other route | 15:03 |
rick_h_ | skip ingestion, use the store upload command | 15:03 |
mbruzek | rick_h_: This is a personal namespace bundle, we haven't promulagated it | 15:03 |
mbruzek | rick_h_: I don't understand the difference | 15:03 |
rick_h_ | mbruzek: after the format it fixed | 15:03 |
rick_h_ | juju store upload talks directly to the charmstore and doesn't involve synciung with charmworld or pulling in updates from bzr | 15:03 |
lazypower | rick_h_, i recall this, we were going ot do store upload for some of the bundles in k8's repo | 17:31 |
lazypower | but we haven't got there yet | 17:31 |
rick_h_ | lazypower: ah ok | 17:31 |
lazypower | I'll propose a fix for the plumgrid bundle before i EOD today, and get that fixed up | 17:32 |
lazypower | moving forward, would you prefer to get the exercising in on the upload bits? | 17:32 |
rick_h_ | lazypower: you poke at the spec for the future upload work? | 17:32 |
rick_h_ | lazypower: hopefully will provide good things for your folks | 17:32 |
rick_h_ | lazypower: makes no difference to me to be honest | 17:32 |
lazypower | i took a real brief glance over it, i need to dedicate some brain cells to it. I've been sick since i got back from DC :| | 17:32 |
rick_h_ | lazypower: understand | 17:32 |
rick_h_ | thanks for taking some time | 17:32 |
lazypower | i'll give it more brain power today and land comments | 17:32 |
lazypower | we've been grasping at straws trying to get the CWR working for k8's bundles presently | 17:33 |
rick_h_ | cwr? | 17:33 |
lazypower | our github => launchpad promotion process has left us with some inconsistencies | 17:33 |
lazypower | Yeah, Cloud Weather Report | 17:33 |
rick_h_ | ah ok | 17:33 |
lazypower | so, in short, store upload decoupling cant come fast enough if this is our chosen path forward :) | 17:33 |
lazypower | stick our upload logic in CI and call it a day for things being out of sync. | 17:34 |
rick_h_ | lazypower: yes, that's why I'm pushing for feedback. Some of the team have already started on small bits and we want to plow forward and get it out to folks to use when we kill v3/charmworld at the end of hte year | 17:34 |
lazypower | i'm +1 for that | 17:35 |
lazypower | now if i remember orrectly | 17:35 |
lazypower | once i store upload, that branch is pretty much forever tied to that new workload | 17:35 |
lazypower | s/workload/workflow | 17:35 |
rick_h_ | huh? | 17:35 |
rick_h_ | oh hmm...pretty much | 17:36 |
lazypower | once the store finds it in its own data store, it no longer looks @ charmworld for data, right? | 17:36 |
lazypower | i think mbruzek and I will be +1 for adopting that workflow and killing off the porting between VCS tools. | 17:36 |
lazypower | it skips review process though | 17:37 |
rick_h_ | lazypower: right | 17:37 |
rick_h_ | lazypower: on both accounts | 17:37 |
lazypower | mbruzek, any thoughts/comments on this? ^ | 17:37 |
lazypower | shall we be guinea pigs on this and reduce our cost of ownership to port between GH and LP? | 17:37 |
rick_h_ | lazypower: that's the issue is getting the review queue work pulled in. Talked to marcoceppi and tvansteenburgh on that a bit yesterday. | 17:37 |
lazypower | yeah, it was a topic at our standup too | 17:37 |
rick_h_ | so increase in manual watching over review/udpates | 17:37 |
lazypower | i think we can gate through CI to do this though | 17:38 |
lazypower | and have the CI process be our source of truth until the new revq lands | 17:38 |
lazypower | its ~ a day of engineering to land that work when coupled with tvansteenburgh's jenkins scripts | 17:39 |
rick_h_ | lazypower: ok, let me know what we can do to help | 17:40 |
lazypower | i think i've identified a blocker though: we have no way to push our branches from namespace to ~recommended | 17:40 |
lazypower | and what this has in terms of implications on CWR/CI | 17:41 |
rick_h_ | lazypower: ah, we have a promulgation command/api call | 17:41 |
rick_h_ | lazypower: might have to check how to get you able to use it | 17:41 |
lazypower | ok | 17:41 |
lazypower | and CI/CWR? i guess we need to do a litmus charm/bundle and see whats missing? | 17:42 |
rick_h_ | lazypower: shoot an email to urulama rogpeppe and myself | 17:42 |
lazypower | will do | 17:42 |
rick_h_ | we'll try to unblock you | 17:42 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!