=== andrewsmedina_ is now known as andrewsmedina [02:35] HTTPS okay to avoid hash check? [03:03] marcoceppi: as long as you are verifying the certificate [03:03] marcoceppi: which wget does by default [03:03] what about git clone via GIT+SSH? [03:04] Would need to have the ssh server key in the charm. [03:04] But then, yes [03:04] cool, cool [03:04] thanks [03:04] as long as you have a high degree of confidence that the content is the content you intended [03:05] Like my highschool self, I'm low on confidence :) [03:06] marcoceppi: interesting .. who owns the S3 bucket MinecraftDownload ? [03:06] That's Mojang's official bucket [03:06] marcoceppi: ok, so if they ever deleted it, could someone else pwn it? [03:07] marcoceppi: I think thats just "something to look out for" and not a reason to be worried [03:07] SpamapS: that's a good question though, I'm not sure [03:08] I'll try that out with my personal vs work AWS accounts, see if names are locked in forever or not [03:09] marcoceppi: seems like they should be mutable if you can delete them [03:09] Would make sense [03:13] marcoceppi: reviewed proposed change [03:14] SpamapS: thanks! I understand calling stop and start, but why install? Should I have the check in install to see if it's been "installed" before there? [03:14] SpamapS, are you in santa clara? [03:14] * hazmat is as well [03:22] hazmat: haha yes === flaviamissi_ is now known as flaviamissi === fenris is now known as Guest47264 [03:22] hazmat: actually I'm at the SJC airport [03:22] hazmat: didn't realize pycon was here as well, or I'd have tried to get in touch w/ you guys [03:22] hazmat: I did bump into some canonical peeps at the hyatt before realizing charm school was at the hilton across the street. ;) [03:23] SpamapS, wanna come over, we're all hanging out sprinting [03:23] i'm sitting next to barryW [03:47] hazmat: I'm in the airport. Depart in 1hr. :-P [03:48] hazmat: just dropped in for a commando attack on the juju users of silicon valley :) [03:49] If I'm updating a charm, and there's no copyright file, should I throw one in there? [03:49] * marcoceppi is not a lawyer [03:50] marcoceppi: do you know for sure the license and copyright status of all the files? [03:50] The only files are the hooks and core charm files [03:51] none of them have copyright headers [03:53] marcoceppi: where did you get those files? [03:53] heh, http://charms.kapilt.com/charms/oneiric/owncloud [03:56] well I-be-damned [03:56] yes, yes you be :) [03:57] marcoceppi: can you open a bug in that charm and assign it to koolhead17 ? [03:57] yeah [03:57] should not have been promulgated [03:57] anyway.. since I'm damned.. I think I"ll go get a mojo burger [03:57] * marcoceppi looks sternly at SpamapS for not being perfect [03:58] * SpamapS looks hard a-stern .. [04:00] bug opened [04:00] Going to head off, have a safe flight! [05:07] marcoceppi, hey there [05:07] hey SpamapS [05:45] <_mup_> Bug #953732 was filed: Destroying a maas environment results in a traceback < https://launchpad.net/bugs/953732 > === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === asavu_ is now known as asavu [12:45] I'm getting the following error when I try to start a local charm: [12:45] 2012-03-13 08:44:12,459 INFO Searching for charm [12:45] Invalid value for cache: False [12:45] ...well, I'll google first... [12:46] Looks similar to https://bugs.launchpad.net/juju/+bug/952355 [12:46] <_mup_> Bug #952355: Backward incompatible change: existing env boolean values cause total failure < https://launchpad.net/bugs/952355 > [12:46] gary_poster: Which charm? [12:47] marcoceppi, buildbot-master (which we wrote) [12:47] looks like something to do with a config value not being a string? [12:47] gary_poster, yep [12:47] ok, looking, thanks [12:47] basically if juju thinks it might be a boolean it won't let the config value be string [12:48] :-/ [12:48] I hit this in several of my charms when running against juju PPA [12:48] its also annoying that juju will fail if ANY charm in your repository has this issue - not just the one you are trying to deploy... [12:48] (that took me a while to figure out) [12:48] oof, wow. jamespage, is there an easy workaround for now? [12:48] jamespage: Oh, that's weird [12:48] I mean, even in changing config [12:49] * gary_poster goes to look for explicit yaml string syntax, if there is such a thing [12:49] gary_poster, switch anything that has values 'True' or 'False' to type: boolean [12:49] rather than string [12:50] same applies to uncapitalized versions of those literals [12:51] or don't use true/false [12:53] jamespage, ack, thank you. I don't think it's even in my charms--which charms did you find with the problem? [12:53] gary_poster, cassandra had it but I've fixed that one [12:54] I suppose I could get rid of all the ones I don't currently use...and try to get newer versions of all of them [12:54] gary_poster, might make sense for the time being [12:54] jamespage, ack, thanks again [12:54] I'm hoping that we might get a fix to sort out backwards compat [12:54] its a PITA [12:54] gary_poster, np [12:55] agreed on PITA === almaisan-away is now known as al-maisan [13:14] jamespage: ah glad I just checked in here, was about to file a bug report for the same thing :) [13:15] bbcmicrocomputer, :-) [13:15] bbcmicrocomputer, both hadoop and hbase suffered from this change [13:15] I've fixed both up for precise [13:15] jamespage: ah cool, I'll pull down those updates then [13:16] jamespage: looked to me like juju was just using the YAML parser and allowing it to coerce config objects to any type, hence not allowed boolean values as strings [13:42] Good mornings! [13:47] niemeyer: morning [13:48] bbcmicrocomputer: Heya [14:09] R [14:48] marcoceppi: shouldn't that be 'R m8t3' for chat like a pirate day? [15:18] m_3: I suppose... *rubs chin* [15:23] marcoceppi: :) [15:37] m_3: my son was born on talk like a pirate day. Wife was not exactly in the spirit during labor.. but the nurses were all about swabbing the deck and hoisting the mainsail.. ;) [15:37] shiver me timbers, thar be a baby! === al-maisan is now known as almaisan-away [15:46] How can you tell which charms need review? It would appear almost all the new-charms are awaiting feedback from the authors [15:49] marcoceppi, it should be new-charm [15:49] https://bugs.launchpad.net/charms/+bugs?field.tag=new-charm [15:49] right, but they all seem to have been turned over [15:49] check the in progress ones [15:49] k [15:49] incomplete means not ready [15:49] oh dude, new. :) [15:50] we started on diaspora on the big screen yesterday, sort of a collaborative charm writing experience.. it went pretty well. [15:50] marcoceppi, looks like Gearman is ready for review [15:50] and disaspora is.. nightmarish to setup [15:51] it looked so not fun [15:51] negronjl said it felt a lot like cloudfoundry [15:53] marcoceppi: new-charm tag _and_ status of "new" or "fix committed" [15:53] m_3 jcastro thanks! [15:53] SpamapS: yay! I totally wanna see diaspora working [15:56] m_3: 18 tons of ruby gems to install. :-( [15:56] I can't believe rubygems, pear, and pecl, don't use any kind of crypto to secure their downloads [15:56] marcoceppi: been trying to keep it as a conversation... review and update status to incomplete or in progress... then they set status to fix committed for a nother round of review [15:57] SpamapS: use bundler... there's a Gemfile [15:57] m_3: makes sense, easy enough to track [15:58] m_3: bundler is what we used [15:58] SpamapS: gem won't check certs on https sources? [15:58] m_3: gem doesn't use https [15:58] rhuh? [15:59] it didn't last year when I was patching it for some other bugs and I got curious [16:01] I'd have to look... thought most sources were specified via https github urls [16:01] m_3: and where does it get those source URLs? [16:01] you can set that in the gemfile [16:02] don't know for the "standard" rubygems.org gems tho [16:02] m_3: ok, so I'm more talking about when I type 'gem install foo' [16:02] right... dunno for the stds [16:03] how'd charmschoolsv go? [16:03] how many peeps were working along -vs- osmoting? [16:04] SpamapS, m_3: I'll keep working on diaspora. I'll commit the work soon. It's a pain in the neck but, I like a challenge :) [16:05] negronjl: cool! [16:05] negronjl: I'll try to drum up volunteers to maintain it going forward [16:06] m_3: that would be great ! It's a complicated mess of gems, ruby packages and all other sorts of nonesense [16:06] so.. typical charm :) [16:07] negronjl: yeah, hitting a ruby conf later this week... I'll try to drum up help [16:07] m_3: nice! [16:11] negronjl: were there lots of peeps writing stuff? [16:12] negronjl: or was it mostly people wanting to customize/use existing charms? [16:12] m_3: Some of the guys kept asking very intelligent questions ( outside of what we were talking about ) which tells me that they were already investigating the use of juju ( and charms ) for their own projects. [16:12] negronjl: awesome [16:12] m_3: We had some guys that were looking at the existing charms ( couchbase and cassandra come to mind ) to see how they can better the charm. [16:13] dude, that's great [16:13] m_3: The DataStax guy in particular was already looking for ways to make cassandra better .... that part was awesome! [16:13] how'd we publicize this event? [16:13] sounds like you had a great group of people show up [16:13] m_3: Some of the business guys here sent email out to their customers ... that was it [16:14] ok, gotcha [16:14] m_3: indeed ... very smart and technical people. [16:14] cool... thanks for the update [17:08] SpamapS, are you guys back from travel? like should we have a linkup today? [17:09] jcastro: yeah it was a 1 day thing :) === tobin is now known as joshuaBRB [17:09] jcastro: I have a team call.. then need to catch up on email. So, probably 90min.. but then yeah I want to chat w/ you and Juan [17:09] yeah whenevs [17:09] negronjl: can you join us from the land of barristas and lattes? [17:09] just the usual charm team hang out, we'll snag mims [17:10] SpamapS: here [17:10] SpamapS: can you guys give me the Hangout URL ? [17:11] negronjl: when we start it.. 90 min [17:11] SpamapS: yeah ... I jumped the gun there didn't I :/ [17:13] negronjl: perhaps try thinking about baseball.. [17:20] ewww [17:43] so how exactly does the new constraints stuff work? any good docs on usage yet? === fjlacoste is now known as flacoste [17:44] jamespage: I believe there's a spec [17:45] hm, not in the trees anywhere.. weird [17:45] SpamapS, http://bazaar.launchpad.net/~juju/juju/trunk/view/head:/docs/source/drafts/placement.rst [17:45] ? [17:47] jamespage: actually yes [17:47] that does seem to mention it [17:47] though light on implementation details like how to use them [17:48] SpamapS, agreed [17:50] jamespage: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=589386 [17:50] jamespage: even weirder than needing a restart.. [17:50] you need *two* reloads [17:52] * jamespage faceplants [17:52] why oh why oh why [17:52] jamespage: pretty odd. The recommended solution is to do 'restart' on initial install, and 'force-reload' on upgrades [17:53] I think I'll submit a patch for that to Debian and fix it in our packages once we get some clarity on 5.4.0 [17:53] SpamapS, OK - sounds sensible [17:54] jcastro: G+? [17:54] negronjl: ? [17:54] m_3: ? [17:54] SpamapS: here [17:55] actually let me reboot to see if I get my video back [17:55] SpamapS, give me 5! [18:03] hrm.. [18:03] * SpamapS tries new kernel [18:10] * SpamapS rejoices [18:10] invites out [18:13] * SpamapS lurves audio hell [18:20] jcastro: ?? [18:34] SpamapS: in a meeting... [18:51] SpamapS negronjl jcastro : damn... just tried to join [18:52] dohhhh haha [18:52] miss anything [18:52] ? [18:52] m_3: too funny [18:52] jcastro did a triple lindy [18:52] m_3: maybe we left 'cause we saw you trying to join [18:52] we were hanging out talking about juju arm tests [18:52] no way! [18:52] I love the triple lindy [19:22] <_mup_> Bug #954350 was filed: juju bootstrap doesn't ensure zookeeper is installed < https://launchpad.net/bugs/954350 > === koolhead17|away is now known as koolhead11 [19:32] SpamapS, marcoceppi if anyone can update the charm repo with my latest commit, i have added the copyright file!! :) [19:32] koolhead11: I saw that! I've got an update that will put it in place [19:33] marcoceppi, cool!! cheers!! === Furao_ is now known as Furao === joshuaBRB is now known as tobin [21:08] m_3, it feels like the timeout on the charmrunner service watching needs to be increased to account for the container creation [21:09] hazmat: hey, ... lxc-start-ephemeral.. why can't we use that in juju instead of lxc-start? [21:09] or rather, instead of lxc-clone+lxc-start ? [21:09] SpamapS, we do lxc-clone [21:09] Right, ephemeral is now using overlayfs .. allowing all containers to share the same VFS cache [21:09] on an ssd its pretty fast to get a new container for a new unit [21:10] not about speed [21:10] the charmrunner is reusing the env, so the template container is still around [21:10] cache pressure is intense w/ 7 containers running [21:10] think .. laptop w/ 4GB of RAM.. spinning up openstack on that pushes it into swap. [21:11] doc it hurts when i kick myself ;-0 [21:11] SpamapS, is that overlay persistent? [21:11] hazmat: don't know. [21:11] perhaps? [21:11] hazmat: aufs was persistent.. so I presume overlayfs (which is supposed to be a replacement) is too [21:12] i guess i should play around with it [21:12] hazmat: tho "lxc-start-ephemeral" suggests that the containers maybe are not. [21:13] hazmat: I wonder.. maybe we can mod lxc-clone to use it somehow [21:13] * SpamapS wonders why we're re-inventing schroot in lxc's tools [21:13] SpamapS, yeah.. its creating the overlay in /tmp [21:26] hazmat: ok, so in theory, we could just change our usage of 'lxc-clone' to just mount an overlayfs and copy the configs.. [21:27] hazmat: and then push that back upstream to lxc as '--overlayfs' eventually. [21:27] hazmat: one thing that sucks.. overlayfs still doesn't support inotify. :-P [21:28] SpamapS, we could, but really that should be in the ephemeral script to take an overlaydir option [21:28] and then we could just use it.. [21:28] hazmat: except then its not ephemeral :) [21:29] SpamapS, the only reason its ephemeral is that the script is hardcoded to use a /tmp dir.. else it would be suitable.. but yeah.. point taken [21:29] hazmat: tho because the rootfs is a mount, and not static files on disk, we'd need a pre-start for the container that mounts it again.. I think [21:30] hazmat: or just accept that they are ephemeral make it an environment option [21:30] err.. "and" make it an option [21:50] https://github.com/lg/murder is it possible to use it as charm? [21:51] koolhead11: yes it would be an awesome charm! [21:52] koolhead11: it might morph into a subordinate once that lands [21:52] SpamapS, but how will/can we use it? [21:53] i meant a used case kind :PO [21:55] koolhead11: probably two parts.. one charm for the tracker server, and then one as the client that distributes files. [21:55] koolhead11: dare to dream... if you have a use case.. charm it up.. otherwise.. perhaps leave it for somebody with practical need/experience to make it happen? [21:56] SpamapS, shall i put it as whishlist/charm needed? [22:34] koolhead11: sure. :) [22:39] hazmat: thoughs on bug 952397 .. I saw that you picked it up.. [22:39] <_mup_> Bug #952397: Juju is forcing *ALL* charms in a repo to be perfect to be able to deploy *any* < https://launchpad.net/bugs/952397 > [22:39] thoughs or thoughts .. either is fine ;) [22:40] SpamapS, yeah.. i need to fix up the exception hiearchy to prevent that from reoccuring.. the whole get_charm_from_path returning a charm error if its not a charm screws up the error catching logic to often [22:41] SpamapS, i'm sprinting on mailman right now at pycon (conference day).. but i'll be back in the swing tomorrow [22:41] SpamapS, https://bugs.launchpad.net/charms/+bug/954535 [22:41] <_mup_> Bug #954535: Charm needed: murder < https://launchpad.net/bugs/954535 > [22:41] hazmat: cool I figured it was just an error handling issue [22:42] koolhead11: thanks!