[00:04] Anyone seen issues where cloud-init is setting the hostname to localhost.localdomain? [00:04] The advertised metata-data/hostname seems to be ok! === harlowja is now known as harlowja_away [09:42] ndonegan: That sounds familiar; what OS/cloud are you running? [09:44] Actually, some searching around suggests I might have been thinking of something else. :/ [10:25] Odd_Bloke: Openstack Icehouse (moving to Kilo in the next few months) with Centos 6.5 VMs. [10:26] ndonegan: Is there anything relevant in cloud-init.log? [10:27] Nothing that I can see. I'll get me some caffine and have a look again. [10:27] Caffine taking priority right now :D [10:29] :D [12:14] Absolutely nothing related to update_hostname or friends in the log that I can see. [12:19] Ah! PEBKAC. [12:22] ndonegan: My favourite sort of problem. ;) [12:23] A hacky fix for an older problem has come back to bite. [12:53] Confirmed. Very much PEBKAC. [16:17] smoser: Review and merge of https://code.launchpad.net/~daniel-thewatkins/cloud-init/smartos-v2-metadata/+merge/251775 would be appreciated; it'd be good to get it in to vivid if possible. [16:17] (harlowja_away has already given it a once over) [16:24] Odd_Bloke, ok. will take a look. thanks. [16:25] smoser: Thanks! :) === harlowja_away is now known as harlowja [17:18] Odd_Bloke, is there information on v2 metadata somwhere? or on its differences from v1 ? [17:19] smoser: http://eng.joyent.com/mdata/protocol.html is the specification. [17:19] It's a completely different format from v1. [17:19] The keys stay the same though. [17:20] put a link in code [17:20] ? [17:21] also, i probably dont like how many LOG.info you do. [17:22] there should be well defined criteria for when debug/info/warn are used, but there are not. instead you have smoser saying "probably". [17:25] smoser: I'll switch the info calls to debug. [17:27] smoser: Pushed those fixes up. [17:27] (Including the link to the spec) [17:28] smoser maybe https://wiki.openstack.org/wiki/LoggingStandards#Log_level_definitions should be just what we recommend? [17:30] harlowja, dont you think that'd be better if it said "smoser's judgement call" [17:30] +1 [17:30] lol [17:30] all hail dear leader smoser [17:30] the big thing is console output [17:30] true dat dear leader [17:31] now and in cloud-init v2 , the console should get "OH NO, BAD STUFF" and very few "Yep all went well". [17:31] and then fingerprints and such [17:32] that reminds me, oh where art thou PCManticore person for https://github.com/cloud-init/cloud-init/pull/2 :-/ [17:32] harlowja: that's me. [17:33] sweet [17:33] how goes it [17:33] 'license change GPLv3 to Apache 2.0' [17:33] intersting [17:33] smoser is your boss ok with this :-P [17:33] bossman [17:33] well, I was just waiting for the licensing issue to settle, I should update the pull request this week and hopefully it gets integrated. [17:33] kk [17:34] guess they are settling [17:43] bossman ack'd it. [17:43] harlowja, its actuall "apache 2.0 or GPLv3" dual license [17:43] which , is actually inherent to all apache 2.0 projects [17:43] interesting, didn't know there was an 'OR' [17:43] but it is explicitly stated here. [17:44] intersting [17:44] ie, apache2 is "upstream compatible" with gplv3 [17:44] * harlowja this stuff confuses me, lol [17:44] so you can grab whatever you want that is apache2 and relicense it to someone under gplv3 [17:44] cools i guess, ha [17:44] but if i cna get away with it, i'm going to not explicitly list the GPLv3 info in the file headers. [17:45] but unfortunately, i suspect if a lawyer gets involved, then.. [17:45] i promise not to involve my sister (who is a lawyer), lol [17:45] thank you [17:45] np [17:46] let's keep this between me and u smoser [17:46] no need to involve lawyers, jeez [17:46] jeezsh [17:47] Odd_Bloke, the only other comment... [17:48] you're not sniffing one or the other, just expect that v2 is there? or are you [17:48] smoser: We just expect v2 is there (which is a fair assumption in Joyent). [17:49] is there code we can drop then ? [17:50] smoser: The code which did v1 stuff has already been dropped. [17:51] smoser: We should eventually be able to get rid of the b64- stuff, but doing so now would break anyone using those attributes. [17:52] k. [17:52] ok. so... last thing. [17:52] err.2 quests [17:53] 1 fetch the princess from the castle [17:53] and deliver the bread to the farmer down the road [17:53] a.) regex, compile that once [17:53] harlowja: I'm in [17:53] no quests for u! [17:53] lol [17:53] though we should watch out for the cavetrol [17:53] harlowja, did you come up with that just now ? or are you quoting something that i'm too un-geek-hip to get ? [17:53] if you'd have said "all your base are belong to us" then i'd have known [17:53] smoser u said 2 quests [17:53] i made up 2 quests, lol [17:53] :P [17:54] or "^^vv<><>baba start" [17:54] harmw, it is a plan of mine to build the konima code into cirros somehow. [17:54] koniwhat? [17:55] b.) http://en.wikipedia.org/wiki/Konami_Code [17:55] you never played ikari warriorson NES. [17:55] no man [17:55] I'm to young for that :> [17:56] and I only know 'all your base are belong to us' from bedtime stories [17:56] (i'm with you on that one) [17:56] but ikari warriors was good [17:56] https://youtu.be/wiQmBwwx7_Y [17:56] Odd_Bloke, ^ see 'a' above, and then [17:57] importing konami: +1 [17:57] b. i forgot that. [17:57] then c. 'format' [17:57] i wonder, why [17:57] + return '{0:08x}'.format( [17:57] + binascii.crc32(body.encode('utf-8')) & 0xffffffff) [17:57] rather than [17:58] er... mabye not on that one. [18:03] Odd_Bloke, so the othe rone is bsaically just why: [18:03] + raise JoyentMetadataFetchException( [18:03] + 'Incorrect frame length given ({0} != {1}).'.format( [18:03] + frame_data['length'], len(frame_data['body']))) [18:03] instead of [18:04] raise JoyentMetadataFetchException('Incorrect frame length given (%s != %s)' % (..., ...)) [18:04] i've nver really understood why one is preferable to another [18:05] The two arguments for .format are generally: (a) it supports more sophisticated formatting, and (b) it won't break if you unexpectedly give it a tuple. [18:06] http://stackoverflow.com/questions/5082452/python-string-formatting-vs-format [18:06] yeah. so, i'm not going to nit-pick you on it, but generally id have used what the rest of the code has. [18:07] but fix the regex [18:07] then i'll pull [18:07] Regex fix pushed. [18:07] oh. Odd_Bloke the other question.. [18:07] the LANG thing in tox [18:07] you wanted LANG=C. and that is reported fixed in httpretty [18:08] Yeah, there's an Azure test (which was written based on a real-life failure) that only fails when LANG=C. [18:08] OK, cool. [18:08] I'll have a look at that tomorrow, see if we can use the latest httpretty. [18:08] Gotta run now. :) [18:08] o/ [18:26] Odd_Bloke, its not fixed. i commented at https://github.com/gabrielfalcao/HTTPretty/issues/223 === harlowja is now known as harlowja_away === harlowja_away is now known as harlowja [19:44] hm, think my network is finally settled [19:44] by not using virtio [19:44] and upgrading the router firmware (eg. freeBSD kernel) :P [19:45] -1 for using Fedora 21 as hypervisor instead of old faithfull CentOS 7 [19:46] (though non-virtio throughput sucks quite hard) [19:47] lol, damn centos7 is already old faithful [19:47] wasn't it recently released, lol [19:48] * harmw spend like 4 months debugging this and trying to get shit to work, openstack from scratch plus instances actually working proprly [19:48] and thats without pro-hardware :p [19:49] when do you use virtio-througput? other than storage? [19:49] * kwadronaut facepalms [19:49] good one, well it just sucks going from 930Mbit to 60Mbit [19:50] i'll answer myself: graphics cards. [20:46] smoser: lsblk and lscpu are missing in your cirros buildroot-upgrade branch [20:47] i have vague recollction of that [20:49] hm, can't find the knobs to turn those on though (alteast not after a quick grep through busybox.conf) [20:56] harmw, probably from buildroot