| === Guest69465 is now known as mfisch | ||
| === mfisch is now known as Guest67039 | ||
| === harlowja is now known as harlowja_away | ||
| harmw | smoser: ping | 10:59 |
|---|---|---|
| smoser | harmw, here. whats up? | 13:00 |
| harmw | I'm working on some code to get the proper fbsd interfacename | 13:00 |
| harmw | should be ready soon | 13:01 |
| harmw | can you change the instance hostname in AWS? | 13:01 |
| harmw | btw, don't nuke that instance since I'm debugging on it :p | 13:28 |
| smoser | harmw, you can't change a hostname in aws | 13:35 |
| smoser | i can give you another instance if that'd help. | 13:35 |
| harmw | hmk | 13:38 |
| harmw | so ip-10-0-0-158 is the current hostname? | 13:38 |
| harmw | nah, I think that'll work | 13:38 |
| smoser | harmw, wlel, http://169.254.169.254/latest/meta-data/local-hostname would tell you | 13:43 |
| smoser | but you seem to have no sane tools on that system (curl or wget) | 13:43 |
| harmw | fetch :) | 13:43 |
| smoser | i remember a time before ubuntu whne i liked to play with sharp pointy objects. | 13:43 |
| harmw | :> | 13:43 |
| smoser | rather than things that worked ;) | 13:43 |
| harmw | smoser: https://code.launchpad.net/~harmw/cloud-init/amazon-fbsd-fixes/+merge/237595 | 13:50 |
| harmw | come to think of it, it's only relevant for configdrive setups since that's the only case where the nic is even used :p | 13:51 |
| smoser | hooray! | 13:51 |
| harmw | but nonetheless, it's always better then just using 'some' hardcoded name :p | 13:52 |
| smoser | harmw, oh, cause it reads that in interfaces. | 13:53 |
| smoser | yeah, thats all slated to change | 13:53 |
| smoser | where we'll get info on the mac and find it that way | 13:53 |
| harmw | yea, it should indeed be some higherlevel solution | 13:54 |
| harmw | but for now this'll do the trick | 13:54 |
| harmw | if only hiren_ would wake up | 13:54 |
| harmw | since he actually uses configdrive | 13:54 |
| smoser | harmw, the only coments i have on that merge proposal is that you should in general re-use 're.' | 13:55 |
| smoser | with re.compile | 13:55 |
| smoser | but its fine. | 13:56 |
| smoser | ie: | 13:56 |
| smoser | for line in ... | 13:56 |
| smoser | m = re.match('^\w+', line) | 13:56 |
| smoser | is going to re-compile that regex | 13:56 |
| smoser | where it could have used a cached version | 13:56 |
| harmw | so -always- compiled regexes? | 14:03 |
| harmw | ah, it's inside that for loop | 14:04 |
| harmw | damn | 14:04 |
| smoser | i dont know really about a good rule for when to do it and hwen not to | 14:05 |
| smoser | but clearly in that case on the loop, you're wasting cpu. | 14:05 |
| smoser | dont worry about it. | 14:06 |
| harmw | well in this case I honestly overlooked my own loop | 14:06 |
| harmw | stupid though | 14:06 |
| harmw | else I would've used a compiled regex, since you told that on some other review a while back :) | 14:07 |
| jaxxstorm | is this the right place to discuss cloud-init bugs? | 14:37 |
| harmw | jaxxstorm: yes | 14:38 |
| harmw | bring it on | 14:38 |
| jaxxstorm | I'm using the phone_home module and it's failing because of SSL, but my URL is http:// and on port 80 | 14:39 |
| jaxxstorm | output: | 14:39 |
| jaxxstorm | https://gist.github.com/jaxxstorm/0b2f7b20718bcb462fc6 | 14:40 |
| harmw | could you use paste.ubuntu.com or something please :) | 14:40 |
| harmw | anyway, it says this is the url: http://foreman.hostname.com80/unattended/ | 14:41 |
| jaxxstorm | yeah I've redacted the URL | 14:41 |
| jaxxstorm | but basically it's http://foreman.companyurl.com:80/unattended | 14:41 |
| harmw | ah, and that 80 is deliberatly without prepending :? | 14:41 |
| jaxxstorm | yeah that comes from the foreman | 14:41 |
| harmw | UrlError: [Errno 1] _ssl.c:492: error:14090086:SSL routines:SSL3_GET_SERVER_CERTIFICATE:certificate verify faile | 14:42 |
| harmw | you're using a selfsigned cert? | 14:42 |
| jaxxstorm | well my foreman instance does have a self signed cert | 14:43 |
| harmw | exactly | 14:43 |
| jaxxstorm | but it's also available on port 80 without SSL for callbacks | 14:43 |
| jaxxstorm | so why would it even attempt to use SSL on port 80 with http? | 14:43 |
| harmw | isn't foreman just redirecting (forcing) to use ssl? | 14:43 |
| harmw | c-i won't just start SSL by itself | 14:44 |
| jaxxstorm | if I call the exact same URL with curl it doesn't use SSL | 14:45 |
| jaxxstorm | and doesn't redirect | 14:45 |
| harmw | weird | 14:47 |
| jaxxstorm | yeah | 14:48 |
| jaxxstorm | in any case, is there a way to have it not verify the SSL cert? | 14:48 |
| harmw | that's something smoser would know, though I think it is possible to import a custom CA | 14:48 |
| harmw | which would make c-i accept the certificate aswell | 14:48 |
| jaxxstorm | yeah I saw that module and I may go dow that route if needed | 14:49 |
| harmw | it would be the most elegant solution :) | 14:49 |
| harmw | smoser: could you accept the merger? | 14:52 |
| smoser | merged. thanks. | 14:55 |
| harmw | sweeeet | 14:55 |
| jaxxstorm | okay, never mind, sorry guys | 14:56 |
| jaxxstorm | < Status: 302 Found | 14:56 |
| jaxxstorm | getting redirected to https | 14:56 |
| jaxxstorm | :( | 14:56 |
| harmw | I told you :P | 14:56 |
| harmw | ah, you forgot curl -l | 14:56 |
| harmw | *probably | 14:56 |
| harmw | or actualy.. ah nvm | 14:56 |
| jaxxstorm | I guess I can add the CA cert, but that'll be a pain | 14:57 |
| smoser | jaxxstorm, you should probably be able to insert the ca cert in cloudconfig | 14:59 |
| jaxxstorm | yeah I think I can | 15:00 |
| jaxxstorm | does it rely on a package being present for CentOS? | 15:00 |
| jaxxstorm | how does it add the cert? | 15:00 |
| jaxxstorm | I want to make sure all the required packages are installed before I attempt it | 15:00 |
| smoser | i think it shoudl all be contained in cloud-init (and usage of python-requests) | 15:00 |
| harmw | cloudinit/config/cc_ca_certs.py: distros = ['ubuntu', 'debian'] | 15:01 |
| harmw | would that mean it's not supported on RHEL? | 15:01 |
| jaxxstorm | it seems that way... | 15:03 |
| jaxxstorm | :( | 15:03 |
| harmw | jaxxstorm: open up a bugreport, this is something worth adding I'd say | 15:04 |
| jaxxstorm | would love to, where do I go? | 15:04 |
| harmw | https://bugs.launchpad.net/cloud-init | 15:04 |
| jaxxstorm | done | 15:06 |
| jaxxstorm | https://bugs.launchpad.net/cloud-init/+bug/1378874 | 15:06 |
| harmw | that module is currently using some Debian-specific stuff, but if the logic applies to RHEL it should be fairly easy to get it to support that | 15:10 |
| jaxxstorm | thanks :) | 15:11 |
| jaxxstorm | I'm cheating anyway | 15:12 |
| jaxxstorm | runcmd: /usr/bin/curl -k <%= foreman_url('built') %> | 15:12 |
| jaxxstorm | btw, the app is amazing, has really helped my provisioning process, so thanks! | 15:13 |
| harmw | if foreman is redirecting you, you need a curl -L to follow the 30x | 15:14 |
| jaxxstorm | ah, v. true | 15:18 |
| smoser | jaxxstorm, hm.. i dont knwo. i'm not sure how specific that would be. | 15:36 |
| smoser | harmw, harlowja_away https://github.com/cloud-init | 15:55 |
| smoser | now to get source there. | 15:55 |
| harmw | why? | 16:54 |
| harmw | smoser: what's wrong with LP? | 16:54 |
| harmw | btw, you may nuke that instance now | 17:08 |
| smoser | harmw, ok. i'll kill instance | 17:10 |
| === Guest67039 is now known as mfisch | ||
| smoser | nothing wrong with lp, but want to have something on github for people to find it, and or to contribute. there. some people just more comfortable there. | 17:10 |
| harmw | sounds legit | 17:11 |
| === mfisch is now known as Guest72593 | ||
| JayF | smoser: is the copy in git or bzr going to be canonical? | 17:12 |
| JayF | I guess I could pick a better word | 17:12 |
| harmw | :p | 17:13 |
| JayF | I mean lowercase "c" canonical not uppercase "C" ubuntu | 17:13 |
| JayF | :P | 17:13 |
| smoser | :) | 17:14 |
| smoser | JayF, i dont know. not thought of that yet. | 17:14 |
| JayF | That's the most important thing to know, isn't it? | 17:14 |
| JayF | split brain problem and all that | 17:15 |
| smoser | yeah, its clearly important to know. | 17:18 |
| harmw | +1 for keeping LP master :p | 17:39 |
| JayF | +90001 for making it github | 17:50 |
| JayF | you make it github and all the sudden you'll see more contributions from me | 17:50 |
| JayF | that's for sure | 17:50 |
| === harlowja_away is now known as harlowja | ||
| harlowja | holy crap, github, woah | 18:15 |
| harlowja | although the PR mechanism sux still | 18:15 |
| harlowja | *imho | 18:15 |
| gholms | On, the bright side, it means not having to use bzr. | 18:52 |
| kwadronaut | heh | 18:53 |
| kwadronaut | only the maintainer(s) have to take care of keeping everything in sync | 18:53 |
| gholms | Yep | 18:55 |
| kwadronaut | for git-git you can use hooks or send a message to github asking to mirror it (which means they'll pull in changes twice a day or so and it's more obvious for users that it's not the canonical one) | 18:57 |
| kwadronaut | there's both bzr-git and git-bzr fwiw. besides that, i dislike both platforms. | 18:58 |
| kwadronaut | (if we're shedding bikes, i'm taking part and now i'll run again) | 18:58 |
| JayF | I paint my bikeshed git of any kind | 19:02 |
| JayF | smoser: another option would be to put cloud-init in stackforge/ and use openstack processes | 19:02 |
| JayF | smoser: I have no idea how you'd feel about that, but you'd get some quantity of testing infra for "free" | 19:03 |
| JayF | smoser: plus you'd be more tightly integrated with one of your biggest users | 19:03 |
| harmw | stackforge sounds cool | 19:41 |
| smoser | JayF, this is true, and i've considered that before. but i'm not tied to openstack, and do not really want to be. | 19:49 |
| harlowja | i'd have to lean towards not also, i've got about 3 stackforge projects and an oslo one, so i've been through it, i'm not sure its buying much for cloud-init that github + travis couldn't buy (i do personally like the gerrit that openstack has over pull requests, but this is a managable dislike) | 19:51 |
| harlowja | although the stackforge one could have an interesting connection into the rest of openstack if that stackforge repo somehow got plugged into the image-building | 19:52 |
| === Guest72593 is now known as mfisch | ||
| harlowja | would be nice to have a new image-built when cloud-init review happens, and tested accordingly, thats like the true functional test that would be nice to have automated | 19:53 |
| === mfisch is now known as Guest6813 | ||
| harlowja | but if said thing doesn't exist ( JayF might know more about the image building stuffs that could be possible) then i'm not sure it matters, github, stackforge... | 19:54 |
| === Guest6813 is now known as mfisch | ||
| harlowja | i've had https://github.com/yahoo/Zake working with travis and for a project without ton of activity its worked out ok | 19:55 |
| smoser | harlowja, i want an easy "patch image with cloud-init" for sure. | 20:02 |
| smoser | but i dont think building the image from scratch is useful. | 20:02 |
| harlowja | well scratch or whatever other mechanism that i believe the tripleo folks/ironic people are using | 20:02 |
| smoser | and 'mount-image-callback' or libguestfs makes patching it in trivial. | 20:02 |
| harlowja | sure | 20:02 |
| smoser | http://ubuntu-smoser.blogspot.com/2014/08/mount-image-callback-easily-modify.html | 20:03 |
| harlowja | be nice to have whatever that is part of the review 'check' pipeline | 20:03 |
| smoser | this is ture, but openstack would specifically be better off using a "stable" cloud-init for testing (ie, grabbing latest ubuntu image). | 20:03 |
| smoser | and probably even latest image of stable release. | 20:03 |
| harlowja | :) | 20:04 |
| harlowja | so the oslo program has done something sorta neat for that | 20:04 |
| harlowja | it has a check pipeline that checks the existing (stable) and from source | 20:04 |
| harlowja | so u can get early warnings from the source 'check' and still not affect users of 'stable' | 20:04 |
| harlowja | ex @ https://review.openstack.org/#/c/126791/ (the gate-*-src-taskflow are testing this source version) | 20:05 |
| harlowja | while afaik the other users of taskflow are not using that source version (but the last release thats avaialble) | 20:05 |
| harlowja | so this could do something similar | 20:07 |
| smoser | hm.. | 20:14 |
| smoser | so that would seem like a 'nice to have' | 20:14 |
| smoser | but stackforge would not seem like a requirement for that. | 20:14 |
| smoser | in that they have "sources" that live other places. | 20:15 |
| smoser | such as cloud-images.ubuntu.com or fedora.mirror.org/foo | 20:15 |
| harlowja | smoser agreed | 21:03 |
| harlowja | its not a requirement, but could be simpler to make happen (emphasis on could) | 21:04 |
| harlowja | since we already have people here in this channel that work with openstack and the infra there | 21:04 |
| harmw | openstack? what's that | 21:26 |
| harmw | sounds scary | 21:26 |
| JayF | I suspect if you moved cloud-init into stackforge/ you could easily convince people to help you get some of this testing working | 21:32 |
| JayF | because of the large number of crossover between openstack devs/opers and cloud-init users | 21:32 |
| JayF | I'm not saying *I* would (I owe this project other work already), but I'm saying people would :) | 21:32 |
| === shardy is now known as shardy_z | ||
| harlowja | JayF so that means u would? | 21:44 |
| harlowja | lol | 21:44 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!