=== gfidente|afk is now known as gfidente
boolmanhey guys, How do I create an account with the password in plain text? I tried this http://pastebin.com/HkVAZT24 but it doesnt work11:53
boolmani also tried in quotes, eg 'installer'11:53
boolmanohh I needed to set lock_passwd: False11:58
smoserboolman, yeah.15:04
larskssmoser: any thoughts on when we might see 0.7.7?17:38
larsksOh, huh, ubuntu ships a package based on a pre-release checkout it looks like.17:39
smoserlarsks, i acknowledge we need one.17:40
smoseri'd 'like to get a list of bugs that we should target, and just start killing them.17:40
smoserso long...17:40
smosersince last release.17:40
larsksYeah.  I'm carrying a bunch of backports in the RHEL package and would love to stop doing that :)17:40
larsksI am happy to help squash bugs if you have a list of release blockers..17:41
smoserlarsks, are you subscribed to https://launchpad.net/~cloud-init17:42
larsksI am now :)17:42
smoserbah. harlowja not here.17:43
smoserholey moley. speak of the devil17:44
smoserharlowja, i said your name literally not 45 seconds before you joined17:44
harlowjai have risen17:44
harlowjafrom a conference in arizon17:44
larsksharlowja is in your irc servers, watching your messages...17:44
smoserlarsks, is asking fora release.17:44
harlowjahas it reached 2 years yet?17:45
smoseri think harlowja is outside my window!17:45
harlowjawe can only release every 2 years17:45
harlowjado we want to have that CI working before  a release17:45
harlowjamake sure things are tidy before that?17:45
harlowja(might be nice to have that same CI activate the building-packages code, ensuring it works)17:46
harlowjathen sure?17:46
smoserthis is true...17:46
smoserwell i'd like to get a release to have something.17:46
smoserand then we can add more releases.17:46
smoserhopefully less than 2 years from now.17:46
harlowjasooo the other stuff i was liking to do was to have a cloud.cfg that was for fedora17:47
smoserthe rsyslog thing is obnoxious for sure.17:47
harlowjaya, that to17:47
larsksharlowja: you mean, a fedora compatible cloud.cfg in the distribution?17:48
smoserso i'd like to do a round of bug triage, and we can target some to a 0.7.8 release.17:48
harlowjain the tree larsks17:48
smoserand then we can go through and move them out if we want17:48
harlowjaat least for people to look at, vs having to find the source rpm, know how to extract it ...17:48
smoserer... i guess 0.7.617:48
smoserer... i guess 0.7.717:48
harlowjai mean we have a brpm tool in tree, but if it uses the default cloud.cfg, that probably not gonna work out the best17:48
smoseri'm so used to seeing 0.7.7~ that i have started to assume that that was the releas.e17:48
larsksOkay. Any objection to just grabbing what's currentinly in the fedora packages and committing it?17:48
harlowjalarsks not from me17:49
harlowjathere is a syslog patch in the fedora package as well17:49
harlowjawhich goes back to how its weird and i'd almost like file logging to be the default in cloud-init17:49
harlowja(providing a logging config that is basic by default, and known to work)17:49
larsksThe only fedora rsyslog related patch appears to be a one-liner changing the match critera in the rsyslog config?17:50
harlowjaya, so i've tried that one, it doesn't seem to work, lol17:51
harlowjaat least when i was using it17:51
larsksHaving said that, I think the logging is a little mucked up right now anyway.  I would rather we were just spitting everything to stdout/stderr and let the system take care of logging it.17:51
harlowjalarsks ya, or just dump it to a file (by default)17:51
harlowjawhich we can do with a 1 line change17:52
harlowjai'd rather not have mucked up logging, but simple stuff that should just work, lol17:53
smoseralright. so very informal. lets say we do a release of 0.7.7 2 weeks from today.17:53
harlowja2 years from today17:53
harlowjacan we arrange to do it on smoser bday17:54
harlowjawhen is your bday17:54
larsksLooking at cloud.cfg...it would be nice if the distribution-specific parts were in something like cloud.cfg.d/distro.cfg or something, so that we could have a common set of modules enabled by default across distribution.  Does that make sense at all?  Right now the cloud_*_modules sections differ substantially between stock/freebsd/fedora...17:57
larsksMaybe that gets too complicated.  I dunno.17:57
smoserwell, for more complexity there, i'd rather hav the modules declare supported distros and improve how the list is done and such.17:58
harlowjasmoser right, we have basic support for that already17:59
harlowjait more though just results in a 'warning' if the module supports an operating system and its being ran on a different one18:00
harlowjaso perhaps just further stuff there18:00
harlowjathen we can just 'list all modules'18:00
harlowjaand let the runtime figure out what should/shouldn't work18:00
harlowja3https://code.launchpad.net/~harlowja/cloud-init/cloud-init-file-logging/+merge/300798 is the logging change18:01
harlowjaif we desire it18:01
harlowja* https://code.launchpad.net/~harlowja/cloud-init/cloud-init-file-logging/+merge/30079818:01
harlowjai don't think most people know that http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/stages.py#L808 exists :-P18:03
smoserok. harlowja larsks i just added a team name cloud-init-bugcontrol18:03
larsksI am fine with the idea of just logging to a file by default.  Can we get file logging directly without spawning a new process like that?18:03
smoserand makde it the "bug supervisor" for http://launchpad.net/cloud-init bugs18:03
harlowjalarsks spawn new process?18:03
larsksharlowja: output: {all: '| tee -a /var/log/cloud-init-output.log'}18:04
smoseryou should be able to move state of bugs now i think18:04
harlowjathat's all in smoser realm for that one :-P18:04
smoserlarsks, log to a file without a process ?18:04
smoseryou just want to avoid the tee?18:04
larskssmoser: log to a file just use the python logging api, rather than having to pipe output to something.18:04
harlowjathat one i think is for 'print'18:05
larsksThat just strikes me as weird.  But whatever works...18:05
smoseroutput: handles any subprocesses18:05
larsksOh right, makes sense.18:05
harlowjaright, also subprocesses18:05
larsksNeeeeeever mind.18:05
smosercloud-init makes its standard out (and inherited) go to places described in output:18:05
harlowjawe can alter that to adjust stdout and stderr18:05
harlowjathat could avoid the tee18:05
smoserthe python logging is configured in 05_logging.cfg18:06
harlowjalarsks  i mean if u really want, we could likely drop the tee18:06
larsksharlowja: Eh, never mind.18:06
larsksI'd rather get a release :)18:06
harlowja200 years18:07
harlowjasmoser does https://gist.github.com/harlowja/e2333dd3ed219b53ef44671185177c3e actually skip anything?18:10
harlowjaeven though things get accumulated in forced and skipped there, those lists only seem to get used for logging?18:11
harlowja(or maybe i'm confused, ha)18:11
smoserharlowja, you would appear to be correct :)18:14
smoserlarsks, can you try something for me ?18:15
smoserso to get a realease... i added you and harllwo to bug control team18:15
smoserand you should be able to now target things you think valid to 0.7.8 milestone18:16
smoserfor example... https://bugs.launchpad.net/cloud-init/+bug/1603830 seems reasonable18:16
smosercan you just try to confirm that (mark it confirmed, it sure seems legit) and then target it to milestone ?18:16
larsksSure.  Let me try...18:16
smoseri can spedn some time going through bugs and targetting them to 0.7.818:16
smoserand woudl appreciate it if you and harlowja did the same.18:17
harlowjayes sir boss18:17
smoserrharper, is also in that group18:17
smoserand powersj18:17
larsksThat seemed to work.18:17
rharpersmoser: ok18:17
smoseralright then.. we have the start of a bug list18:17
smoser https://launchpad.net/cloud-init/+milestone/0.7.718:18
rharper0.7.7 or 0.7.8 ?18:18
harlowjacloud-init 0.7.7 "been-to-long"18:18
smoserrharper is in the same camp as me and has just seen 0.7.7 for so long that he thought that was released :)18:18
smoserthe ubuntu versions are 0.7.7~bzrREVNO18:18
smoserwhere '~' in apt speak means less than18:19
smoserdpkg speak i guess18:19
harlowjawhats ubuntu18:20
smoserits one of those free linux things18:23
smoserkind of like centos18:23
harlowjaexcept newer?18:23
harlowjahttps://code.launchpad.net/~harlowja/cloud-init/cloud-init-really-skip-baddies if u want it :-P18:26
harlowjathat will actually skip modules, lol18:26
smoseris worked_distros close to sane ?18:27
harlowjain the various modules?18:28
harlowjai think the redhat/yum ones probably need updating18:29
smoserdoes larsks have a list that he runs ?18:29
harlowjaso no, imho, not all the modules very well use that18:29
harlowjasome do, some don't, ha18:29
harlowjaones like cc_rhn_subscription should18:29
smoserif its in their default list, i'd mark it as working18:29
harlowjaor cc_yum18:29
harlowjacc_lxd also should probably have debian...18:30
harlowjaso there probably should be an audit...18:30
larskssmoser: A list of modules?  Honestly, I'm not sure that anyone has paid much attention to the cloud_*_modules sections of the config.  Hence my earlier comments about moving to some sort of single list or something...18:30
larsksThat is, there is a config we are using, and it includes some things intentionally, like cc_rhn_subscription, but other stuff just because it was already there.18:31
harlowjalarsks so just some background18:31
harlowjaeach module can optionally have a attribute 'distros' or 'osfamily'18:32
harlowjathe cloudinit runtime (the thing that executes those modules) will look for these attributes and potentially skip if not on a supporting distro18:32
harlowjaby default if a module doesn't list that attribute, its assumed to 'just work'18:32
harlowjasome modules do list it (but without https://code.launchpad.net/~harlowja/cloud-init/cloud-init-really-skip-baddies things aren't actually skipped anyway)18:33
harlowjaso we can have a single list, but we have to list those attributes correctly18:33
harlowjaand then the runtime itself will handle skipping on unsupported distros18:33
harlowjajust we have to consistently mark 'distros = ['ubuntu', 'debian']' in modules18:34
harlowjaso that could == single list if we use it correctly18:35
larsksFair enough.  Fwiw, I would love to see an is_available() method or something that could perform arbitrary runtime checks (e.g., cc_puppet could check if puppet were available, etc), rather than checking against distro names (like, what if someone were running an rpm based distribution with apt?)18:36
larsksmissed him by this much...18:36
larsksharlowja: Repeating myself re:the distros var, because you quit at just the wrong moment: Fwiw, I would love to see an is_available() method or something that could perform arbitrary runtime checks (e.g., cc_puppet could check if puppet were available, etc), rather than checking against distro names (like, what if someone were running an rpm based distribution with apt?)18:39
larsksBut obviously not right now :)18:39
harlowjalarsks why not right now?18:39
harlowjaits not that hard18:39
mgagnelooks like 'name' is optional in links http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/sources/helpers/openstack.py#L547 but later is mandatory: http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/net/network_state.py#L28418:39
larsksWell, if we're trying for a 0.7.7 release, something that introduces an entirely new mechanism for like that seems like a bad choice in the short term.18:39
harlowjalarsks well we can at least build in the mechanism right now18:39
larsksAny chance we can work on github, or do I need to learn bzr + lp to submit changes for review?18:40
harlowjai'll make something quick18:40
harlowjasmoser ^^^^^18:40
harlowjasmoser is mr.moveittogit18:40
smoserharlowja, yeah, lets do it.18:40
harlowjasmoser like by EOD18:40
* larsks yays18:41
harlowjato git18:41
mgagneharlowja: tested bonding + vlans with latest cloud-init and it fails.18:41
smosermgagne, :-(18:41
harlowjamgagne any log u got, any output of input and what it generated?18:41
mgagnefor the reasons I mentioned above18:41
harlowjahehe, let me find the logger for this18:42
mgagnelet me setup network, currently on IPMI18:42
harlowjamgagne can u try https://code.launchpad.net/~harlowja/cloud-init/cloud-init-render-tool18:42
harlowjait should write out files given a config in18:42
mgagneexplain like I'm 5 ?18:42
harlowjagiven network_config.json ---> poof >>> turd18:42
harlowjaor is that el3?18:43
smosermgagne, did that regress ?18:43
smoserharlowja, for git... we have to just rid 'bzr' from everywehre18:48
smoserthat is cloud-init's http://paste.ubuntu.com/20342955/18:48
smoserand then my debian packaging in ubuntu makes use of bzr export and bzr merge upstream18:49
smoserbut that is not a blocker for cloud-inti18:49
smoserfor version sutff. i think we were saying pbr. have to look. i think you had something on that.18:50
mgagnesmoser: never managed to test bond + vlan, only tested "flat" networks (without bond)18:53
mgagnesmoser: testing cloud-init is really a part time job here =)18:53
smosermgagne, right. thats fine. i was just wanting to make sure it wasant regression18:54
mgagneI don't think so18:54
harlowjalarsks https://code.launchpad.net/~harlowja/cloud-init/cloud-init-dynamic-distro-check/+merge/300805 (the basics of that dynamic distro check)18:54
smoseras i was pretty sure you'd verified it working for you.18:54
mgagnethere is nothing to regress at that point18:54
mgagneok, I'm having issue with networking, can't easily fetch logs without SSH. will take more time to debug as it could also be related to our auto networking config.19:03
powersjsmoser, I have an initial job running the python tests now every 12 hours. I'll start looking into the vmtests and how they are run for curtin to see what we can do for cloud-init.19:04
smoserpowersj, yeah, that will be a fair amount of work.19:04
smoserwe shoudl talk some / brainstorm on how to get something reasonable into place.19:05
smoserwe can do a whole lot of test (including non-ubuntu distros) on lxd19:05
mgagneok, that's the best I can do in the short term: http://imgur.com/a/wVuTr19:06
powersjthat's how I've been experimenting is setting up lxd env.19:06
powersjnot sure how to do different archs other than i386 yet though19:06
smoserwell, we have resources for power at least19:07
smoserharlow dead19:09
smoser(summoning worked before)19:10
smoserthis is new to me:19:10
smoser http://paste.ubuntu.com/20345473/19:10
smoserguess which of those lines pep8 doesn't like19:10
smoser E225 missing whitespace around operator19:10
smosermust of come as fallout of https://github.com/PyCQA/pycodestyle/issues/16619:11
larskssmoser: I am looking at the bzr->git stuff in the scripts right now.19:35
smoserlarsks, nice.19:36
smoserlarsks, i think the plan was to use pbr sensible_version and the like19:36
larskssmoser:  does anyone actually use brpm?19:39
smoserharlow does19:40
smoserat least used to frequently.19:40
larsksArg, he's dropped off again.19:40
* larsks is going to buy harlowja an irc proxy for his birthday.19:41
smoserand i'd like to have something like that.. thats what i'd use for some c-i19:41
smoseri use bddeb in development for testing a package.19:41
larsksOkee dokee.19:42
smosermgagne, if you can manage to get the openstack network_json of the failed system that'd be helpful20:06
mgagnesmoser: will try =)20:07
=== gfidente is now known as gfidente|afk
mgagnesmoser: finally made it: http://paste.openstack.org/show/539268/22:47
simaris this channel logged? any publicly available logs?22:52
simaralso how does cloud init keep a track of which modules to run on boot? let's say I write a module that changes the hostname of the machine, would it actively check /etc/hosts on boot each time to know if it's required to run that module or not?22:57
mgagneIIRC there are flags created in /var/lib/cloud/instances/<uuid>/sem for once_per_instance configuration22:58
mgagneotherwise some always run at boot22:58
simargotcha. thanks mgagne that seems to be right23:02
simarAlso if I write a custom module that lives in /etc/cloud/cloud.cfg.d/ how would I call that up from cloud.cfg?23:03
simarso if its xx_modulename.cfg, i'd have to call it as modulename from within cloud.cfg under cloud_final_modules?23:04

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!