bdx | could just be me | 01:50 |
---|---|---|
bdx | is jujucharms.com bugging out right now for everyone? | 01:51 |
dnegreira | ts ok for me | 01:51 |
bdx | safari I get https://imgur.com/a/uJgGT, chrome I get https://imgur.com/a/GYSS6 | 01:52 |
pmatulis | fine here on chrome | 01:54 |
bdx | seem to be getting the same waiting/progress bar on chrome and safari now | 01:54 |
dnegreira | am logged in fine via firefox | 01:54 |
bdx | cool | 01:54 |
bdx | must be a personal issue .. | 01:55 |
bdx | thx thx | 01:55 |
bdx | I try to access https://jujucharms.com and I get redirected to https://jujucharms.com/login/u/jamesbeedy to the spinner/waiting infinity | 01:58 |
bdx | it all works for me in a incognito window | 01:59 |
bdx | must be squirrels in the cache | 01:59 |
ashipika | ping anybody? | 07:39 |
rick_h | ashipika: ? | 12:42 |
ashipika | rick_h are there any restrictions on who can land PRs in core? | 12:42 |
rick_h | ashipika: yes, managed by the github project | 12:42 |
ashipika | rick_h hmm.. it's not letting me land my PR | 12:43 |
rick_h | ashipika: what's your GH username? | 12:43 |
ashipika | rick_h alesstimec | 12:43 |
rick_h | ashipika: hmm, so looks like most everyone has read access so maybe that's not it then. | 12:45 |
ashipika | rick_h https://github.com/juju/juju/pull/8175 | 12:46 |
ashipika | rick_h can't land it with $$merge$$ or :shipit: | 12:46 |
ashipika | rick_h what am i missing? | 12:46 |
rick_h | ashipika: not sure how it's setup :( cmars is a lead which has more rights so maybe check with him when he's on | 12:46 |
ashipika | rick_h will do | 12:46 |
rick_h | ashipika: hmm, yea looks like babbageclunk used $$merge$$ in a previous PR. | 12:47 |
rick_h | ashipika: there was a rule a while ago that your team membership had to be public or the bot wouldn't auth you | 12:47 |
rick_h | ashipika: but in this new UX in GH I can't tell wtf is up with that | 12:47 |
rick_h | ashipika: so honestly not sure what's up | 12:47 |
ashipika | rick_h np.. thanks anyway | 12:47 |
jam | ashipika: so what happened is you did a $$check$$ from a long time ago, which the bot saw but got wedged, and it never replied with a "failed to build" message. | 13:36 |
jam | So all follow up $$merge$$ requests are still treated as part of the original request, which it already handled. | 13:36 |
ashipika | jam a-ha! right.. how do i resolve this? !!build!! again? | 13:37 |
jam | working on it | 13:37 |
jam | if you reply with the magic string that the bot replies with, then the bot will accept future $$merge$$ requests, but I always have to go figure out what the magic string is | 13:37 |
rick_h | jam: ashipika can you just delete the comment with the $$check$$ ? | 13:42 |
jam | rick_h: no. maybe if you deleted the comment from the bot. but hopefully I got the fix in. (got distracted doing a review while trying to find the magic string) | 13:45 |
rick_h | jam: yea, I know originally I used to not $$merge$$ again but delete the bot's acceptance so that it would pick it up as a fresh request | 13:45 |
rick_h | ashipika: if it's bot related like that I'd suggest just gardening the comment thread | 13:45 |
rick_h | the bot's not very smart (at least when I wrote it at the start) | 13:46 |
eriklonroth | Hmm, I can successfully deploy charms behind proxies that are downloaded as "juju deploy ./<charm>.zip", but if I try the same thing with "juju deploy cs:<charm>". It times out. What do I need to do to make the juju-client fetch charms from the store? (or do I just need to mess around more with our authenticated proxy at work? | 15:01 |
rick_h | eriklonroth: the controller needs to have access to the store, not the client. The client will resolve and find the right url, but the controller will fetch the zip file | 15:02 |
rick_h | eriklonroth: is this a maas or openstack or ? | 15:03 |
eriklonroth | Its our first installation of maas we're trying to learn | 15:03 |
rick_h | eriklonroth: can the nodes in maas reach the internet? e.g. does the maas server proxy requests through? | 15:03 |
eriklonroth | The maas nodes can access internet though an unauthenticated proxy | 15:04 |
rick_h | eriklonroth: in my maas at home I setup forwarding on the maas server itself and then the controller (running on the maas node) can reach out and contact the store | 15:04 |
eriklonroth | When you say "forwarding", what does that means in terms of configuration? | 15:05 |
rick_h | eriklonroth: setup NAT on the maas server | 15:06 |
rick_h | eriklonroth: I'm looking for the proxy config settings. I thought they were in the controller-config but I'm not seeing it listed :( https://jujucharms.com/docs/devel/controllers-config | 15:06 |
rick_h | eriklonroth: but if you can bootstrap, the controller machine must have been able to go out and get to the internet for the agents it downloaded...hmm | 15:07 |
rick_h | eriklonroth: I guess try the deploy with --debug and see if that is more useful on where it's getting stuck. | 15:07 |
eriklonroth | NAT:ing isnt the problem for us atm, I think it gotta be related to proxy settings somewhere | 15:07 |
eriklonroth | I'll try mess around more with it. | 15:08 |
rick_h | eriklonroth: but the flow is that the client resolves "juju deploy msyql" into cs:mysql-35 and then the controller is told to go get that | 15:08 |
rick_h | eriklonroth: so that the client doesn't have to download/reupload the charms to the controller | 15:08 |
eriklonroth | OK, so its not the client that first downloads from cs and then uploads it to the controller? | 15:08 |
rick_h | eriklonroth: correct, the client doesn't download it to your client machine. | 15:09 |
eriklonroth | OK, then I'm sure its the controller config - "somewhere" - for the proxy I need to figure out. | 15:09 |
eriklonroth | That will be the exercise for tomorrow. | 15:10 |
eriklonroth | Thanx for the help. | 15:10 |
rick_h | eriklonroth: looks like model level config https://jujucharms.com/docs/devel/models-config | 15:12 |
eriklonroth | I've successfully managed to configure the "default" model, but perhaps its the "controller" model in maas that needs to be setup ? | 15:13 |
eriklonroth | Oh, well, "juju deploy cs:<charm>" isnt' working in the models either. Just from zip-file. | 15:13 |
rick_h | eriklonroth: right, in that case the client is sending the charm up. | 15:13 |
eriklonroth | But apt, pip etc. works | 15:13 |
rick_h | eriklonroth: hmm, well wtf. Yea I guess I'm curious what the --debug output is. | 15:14 |
eriklonroth | 16:14:48 DEBUG httpbakery client.go:244 client do GET https://api.jujucharms.com/charmstore/v5/xenial/ubuntu/meta/any?include=id&include=supported-series&include=published { | 15:15 |
eriklonroth | .... stuck there | 15:15 |
eriklonroth | https://pastebin.com/5MVE1N8F | 15:16 |
eriklonroth | https://pastebin.com/5MVE1N8F | 15:16 |
eriklonroth | Oh well, I have to run. My girfriend will kill me for letting her wait for 15 minutes while I'm messing around with a proxy | 15:18 |
jhobbs | rick_h: do you now if you can use a unit name in a placement directive? | 16:36 |
jhobbs | juju deploy blah --to glance/0 for instance? | 16:36 |
rick_h | jhobbs: thinking | 16:36 |
rick_h | jhobbs: into a container as well? | 16:37 |
rick_h | jhobbs: or just the same machine? | 16:37 |
jhobbs | same machine, whether it's a container or not | 16:37 |
jhobbs | whatever machine is hosting that unit | 16:37 |
jhobbs | hulk smashed with the unit | 16:38 |
jhobbs | hmm seems to work | 16:38 |
rick_h | jhobbs: yea, I don't think so. I was thinking in bundles you can say to colocate with an application name, but you don't get to pick 0/1/2 or such | 16:38 |
jhobbs | no | 16:38 |
jhobbs | it doesn't work | 16:38 |
jhobbs | yeah | 16:38 |
jhobbs | ok | 16:38 |
jhobbs | thanks rick_h | 16:39 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!