| === ejat- is now known as ejat | ||
| niemeyer | Hello jujuers | 13:18 |
|---|---|---|
| smoser | hazmat, i'm here now. it looks to me from your pastebin at http://pastebin.ubuntu.com/918245/ that cloud-init ran the 'start', found the DS, and used it. | 14:16 |
| smoser | but the cloud-config upstart job did not run | 14:16 |
| smoser | which is: | 14:16 |
| smoser | start on (filesystem and started rsyslog) | 14:17 |
| SpamapS | hazmat: re your question about jitsu.. I like the idea of having a 'jitsu' command that is basically just the juju wrapper that already exists.. so it calls juju when jitsu doesn't have a command you pass it. | 14:56 |
| SpamapS | maybe niemeyer should contribute to juju-jitsu, then it would be Brazilian Juju-Jitsu ;) | 14:59 |
| niemeyer | :) | 15:08 |
| imbrandon | jcastro: btw http://brandonholtsclaw.com/sites/default/files/domain-list.pdf I'm thinking cloudexamples.net will be a good place for "sure, just run juju deploy uber wordpress" etc | 15:25 |
| imbrandon | whatduya thunk? | 15:25 |
| gary_poster | hazmat or To Whom It May Concern: my escapades trying to get Juju working on Canonical OpenStack continue. We are now worse off than when I started, but better off than last week. Current status is https://pastebin.canonical.com/63978/ if anyone has any input or suggestions; the text is from the pertinent RT, https://rt.admin.canonical.com//Ticket/Display.html?id=50773 | 15:58 |
| * hazmat joins in the adventure | 16:02 | |
| imbrandon | hrm speaking of OpenStack, marcoceppi did you ever check to see about the s3 compat for hp swift containers ? | 16:06 |
| marcoceppi | imbrandon: I haven't checked since I first tried | 16:07 |
| marcoceppi | too much other stuff going on | 16:07 |
| imbrandon | kk, np , just curious | 16:07 |
| imbrandon | i may dig in and learn the ways of the jedi | 16:07 |
| imbrandon | or try | 16:07 |
| hazmat | gary_poster, what version of juju are you using | 16:09 |
| hazmat | fwereade_, did you try juju with a private cloud with constraints?.. i'm getting errors on ec2-zone | 16:09 |
| gary_poster | hazmat, 0.5+bzr504-1juju4~precise1 | 16:10 |
| * gary_poster gets lunch. biab | 16:13 | |
| hazmat | gary_poster, thanks | 16:13 |
| bkerensa | jcastro: You making juju stickers I hear? | 16:49 |
| jcastro | no | 16:49 |
| jcastro | those are charm school stickers | 16:50 |
| bkerensa | jcastro: ahh | 17:02 |
| bkerensa | jcastro: kees's sbuild juju charm is pretty cool we got to use it when he wrote it | 17:10 |
| hazmat | gary_poster, since the private cloud is behind a firewall, you'll need to login to some machine where the client has access to the cluster. | 17:21 |
| hazmat | gary_poster, attaching the public ip address no longer changes the output of public ip address or public dns address | 17:21 |
| gary_poster | hazmat, that was not the case before, and preventing that restriction (the firewall) is the point of the RT. The .canonistack dns names should be something we can configure once in .ssh/config and then juju wuld just work | 17:22 |
| gary_poster | would | 17:22 |
| hazmat | gary_poster, yes | 17:23 |
| hazmat | it is different now, previously i could just attach a public ip to the bootstrap and work, but that is broken.. if the addresses/hostnames returned we're routable in some fashion that would also fix the problem | 17:23 |
| hazmat | s/were | 17:24 |
| hazmat | where | 17:24 |
| hazmat | ;-) | 17:24 |
| gary_poster | hazmat, :-) cool. So you suspect that if I get the RT fixed (the second bullet point in my summary) then possibly the deployment would also work again (third bullet point)? | 17:25 |
| gary_poster | I was only pessimistic about this because the failure for my third bullet point was seemingly in canonistack itself | 17:25 |
| gary_poster | so it was not a connection failure but...something else, like the security problem you found last time | 17:26 |
| hazmat | gary_poster, hmm | 17:31 |
| hazmat | gary_poster, yeah.. just reviewd the RT in full | 17:31 |
| hazmat | i'd like to think persistence pays off.. | 17:32 |
| gary_poster | :-) | 17:32 |
| hazmat | what i find a little irksome is that previously attaching public ips would change the reported public address | 17:33 |
| hazmat | but that appears to no longer be the case | 17:33 |
| hazmat | and is yet another minor incompatibility with ec2 behavior | 17:33 |
| gary_poster | yeah | 17:33 |
| hazmat | i guess i should submit my patch for euca2ools nova compatibility | 17:34 |
| gary_poster | +1, if that is waiting in the wings, yeah :-) | 17:36 |
| hazmat | looks like its already done.. but un-acted.. https://bugs.launchpad.net/euca2ools/+bug/891889 | 17:36 |
| _mup_ | Bug #891889: euca-describe-addresses fails with missing attribute on Address object when using boto < 2.1 <euca2ools:New> < https://launchpad.net/bugs/891889 > | 17:36 |
| hazmat | gary_poster, i don't really understand elmo's comment on that RT, what is juju supposed to do differently here when it doesn't have access to any accessible endpoint | 17:41 |
| hazmat | ie. re fix juju | 17:41 |
| hazmat | gary_poster, so my recommends in a nutshell would be setting up on the bootstrap instance after logging into it.. or getting the chinstrap gateway setup when they attach a domain suffix to the hostname. | 17:42 |
| gary_poster | hazmat, re fix juju, I think my reply to him addressed his points. There's not something for juju to fix, and there was a reasonable way to accommodate juju (*.canonistack resolving on chinstrap). I will do as you suggest (set up on the bootstrap after logging into it) if I need to do canonistack, but my main goal is to push through the domain suffix thing, because I think that's going to be a win for every person who want | 17:46 |
| gary_poster | s to use juju on canonistack..and we're hoping that's a decent number of people. | 17:46 |
| gary_poster | thanks | 17:46 |
| hazmat | gary_poster, thanks for keeping at it.. i totally agree this process should be easier for folks.. its probably worth a wiki page on how to setup locally that we can direct folks to | 17:54 |
| gary_poster | hazmat, yeah, that's my plan once it works. | 17:55 |
| elmo | gary_poster: what are you talking about? | 18:05 |
| _mup_ | juju/trunk r518 committed by kapil.thangavelu@canonical.com | 18:05 |
| _mup_ | [trivial] fix for private clouds using ec2 api, ec2-zone and instance-type are still valid constraints | 18:05 |
| gary_poster | elmo, RT 50773, your comment https://rt.admin.canonical.com/Ticket/Display.html?id=50773#txn-838466 | 18:06 |
| gary_poster | or actually https://rt.admin.canonical.com/Ticket/Display.html?id=50773#txn-838466 | 18:07 |
| gary_poster | eh same one, sorry, on call | 18:07 |
| _mup_ | Bug #977481 was filed: Support private/public cloud spanning <juju:New> < https://launchpad.net/bugs/977481 > | 19:34 |
| _mup_ | Bug #977528 was filed: Relation membership for child relations should be cached before hook execution <juju:New> < https://launchpad.net/bugs/977528 > | 20:37 |
| _mup_ | juju/relation-hook-commands-spec r9 committed by jim.baker@canonical.com | 20:48 |
| _mup_ | Updated per review comments | 20:48 |
| marcoceppi | Can charms have a "." in the name? | 21:11 |
| bkerensa | marcoceppi: surely jcastro knows | 21:16 |
| bkerensa | :D | 21:16 |
| jcastro | We haven't said one way or the other | 21:16 |
| marcoceppi | I'll just omit the period then | 21:17 |
| jcastro | m_3: hey so, we have a charm on github: https://bugs.launchpad.net/charms/+bug/977552 | 21:18 |
| _mup_ | Bug #977552: Charm Needed: Terraria server <Juju Charms Collection:Confirmed> < https://launchpad.net/bugs/977552 > | 21:18 |
| jcastro | ideas for integrating/mirroring/whatever? | 21:18 |
| marcoceppi | jcastro: you can setup a mirror repo on LP | 21:18 |
| marcoceppi | it pull from git every X mins | 21:18 |
| marcoceppi | I think imbrandon did that for the drupal charm | 21:19 |
| jcastro | I don't think asking people to mirror their stuff in lp will scale | 21:19 |
| m_3 | jcastro: np, with only a few, I can pull manually | 22:33 |
| m_3 | jcastro: still preliminary atm... just treat it as a normal charm submission... have him tag it with new-charm when it's ready for review | 22:37 |
| m_3 | jcastro: the fact that it's on github doesn't matter until after it's accepted into the charm store... then we'll figure out MPs -vs- PRs when we cross that bridge | 22:39 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!