[06:09] * SpamapS tests out an updated juju for precise-proposed [06:25] SpamapS: have you tried that jitsu ? i'm getting all kinda errors with it, python modules missing etc [06:25] wonder if its just that machine is fubar [06:28] imbrandon: pastebin? [06:30] * SpamapS uploads 5 SRU fixes for precise.. hopes rest of SRU team doesn't beat him with a hammer [06:30] http://paste.ubuntu.com/1014277/ [06:31] imbrandon: ahh looks like there are some python library files that have to be installed [06:31] ImportError: No module named aiki.invoker [06:31] imbrandon: confirmed [06:31] okies [06:31] question now is where to install them [06:32] * SpamapS decides libdir [06:32] * imbrandon goes back to breaking other things :) got like 5 projects 99% done [06:32] heh [06:32] i hate that [06:32] imbrandon: lazy is never doing anything you don't want to do.. suck it up ;) [06:34] :) thats the thing i want to do all^most all of them :) [06:34] i just hate haveing 5 at 99% wish i had 2 complete and 2 barely started [06:35] is how i work better , mostly [06:35] but each one had blockers so thus ended up like this for a week [06:35] :) [06:36] gotta go spend most of the after noon witha client later today too, fun fun, i _love_ onsites when there is no real reason for it :) [06:38] imbrandon: people are just complex computers. Hack them like you hack PHP :) [06:41] heh [06:41] never thought about it like that but very true [06:42] so why does jitsu have upgrade-charm and a few other identical sub commands ? [06:42] just curious , dident look at the code to see if they are actually the same or not [06:49] imbrandon: they are meant as wrappers for testing [06:50] imbrandon: sort of a half finished thing [06:52] * SpamapS wills autoreconf and ./configure to go faster [06:55] ahh === al-maisan is now known as almaisan-away [07:00] imbrandon: nearly have run-as-hook working now [07:03] * SpamapS releases that as 0.10 [07:03] nice [07:04] nearly ? you sound like me now [07:04] heh [07:04] no it works [07:04] finally learning the full power of automake/autoconf [07:07] imbrandon: should build in < 10 min [07:08] rockin [07:09] I wish they'd update LP so I could use the latest bzr builder [07:09] has {latest-tag} [07:09] so you can actually have a non-trunk automated build [07:10] heh, like travis builder on github ? hahahahah [07:10] j/k [07:11] imbrandon: you're more than welcome to port all my shit to github for me. :) [07:11] you know i got to raz ya now and then , only a few things i can do it with :) [07:11] though frankly, I'm not all that excited about using a non-open-source service [07:11] would rather we run our own gitolite or something [07:11] heh, i'd be doing good if all mine is, i still have half of it on bitbucket [07:11] that sounds like a GREAT idea to me [07:12] well actually [07:12] I'd rather that LP just grew git [07:12] so people would shut up and stop caring about VCS [07:12] yea but we can use ubuntuwire for what its orig reason for living was again [07:12] just stuff like this untill lp grew it [07:13] qa.ubutnuwire.com came long before qa.u.c :) its how it was modeled :) [07:13] anyhow i checked and there is still 3 or 4 Dell 2650's for ubuntuwire in the canonical dc [07:13] that we could slap it on and do some proof of concept workflow stuff and maybe hurry LP allong [07:14] godamnit launchpad [07:14] 0.10 != 0.1 [07:14] or hell add it ourselfs now its fully open [07:15] i should revamp *.ubuntuwire.com with the new branding, it should all still be auto deploying from the lp bzr branch on push iirc [07:15] huh? [07:15] thinking out loud [07:15] Why wouldn't you just give LP git.. [07:15] ? [07:15] NIH? (I totally get that.. ;) [07:16] heh no [07:16] i would if i thought i coudl pull it off [07:16] not sure a project that big i could do alone tbh [07:16] would want to do a lil planning === Leseb_ is now known as Leseb [07:19] hrm actually there isnt a server component to git like there is bzr, its just bare repos and gitweb ... hrm it MIGHT not be that bad [07:19] mostly just gui stuff [07:19] mostly , not all [07:19] imbrandon: the "server bit" of bzr isn't really all that much. Its just cmds to execute on the server side [07:20] yea but i think thats one reason its so slow too isnt it [07:20] ? [07:20] imbrandon: the harder thing is mapping LP's branch namespacing into git's colocated branches, but I'm sure that would be solvable in a furios night of hacking. [07:20] because omg like bzr or not , you got to admit bzr is SLOW [07:20] imbrandon: "so slow" is so laughable [07:20] most things you do in bzr take 1 - 2 s where as git is 0.5s consistently [07:21] OMG! Bzr! [07:21] SpamapS: push a fresh branch [07:21] imbrandon: Its all relative I guess. :) [07:21] imbrandon: just as fast as git now because of stacking [07:21] and that happens alot in bug fixing and drive by fixes like ubuntun grows alot of [07:21] stacking ? [07:22] yes stacking [07:22] is it actually in ? [07:22] ive not seen it work yet [07:22] still slow here [07:22] lp:~clint-fewbar/ubuntu/precise/juju/mything will automatically stack on top of lp:ubuntu/precise/juju .. any shared commits will not have to be retransferred on push/pull [07:22] stacking has been around for years [07:22] =o [07:22] SpamapS: if the branch is up to date? [07:23] the other "slow" bit is the SSH negotiation [07:23] oh then no, its stil slow as piss then with stacking too [07:23] SpamapS: bs, i can push a clean git over ssh and it takes less than 10s, same branch of omg-wp takes almost 3m on bzr [07:24] imbrandon: thats where bzr's "smart" side comes in because you basically say "I'm at this point in the graph and I want to push to you" and it says "I"m at this point" and then you push whats between. Git just has a map, and you read it real quick, and push what you need. Definitely 1-2 seconds faster, but conceptually identical [07:24] imbrandon: is that git-bzr pushing or real bzr pushing? I'm dubious. Also.. are you stacking on omg-wp somehow? [07:24] real pushing [07:25] the initial push is definitely going to be slower [07:25] because bzr's branch format is far less efficient than git's [07:25] i'll time it real fast if you want, i'll make 2 clean subdirs of omg and git initi and bzr init them toand then run time [07:25] imbrandon: I just did a complete teardown on a brand new System76 :D [07:25] but all subsequent stacked-on branches will be quite fast [07:25] sure, but i do that first push more often than stacks in ubuntu [07:25] as do most drive by contributors [07:27] one sec, i am gonna time it , cuz i'm sure in my head i'm not exagerating here [07:27] will only take a sec anyhow, not like a big test [07:28] imbrandon: why do you do that first push? [07:29] imbrandon: and most drive by's would be checking out the existing branch and pushing a new one stacked on it [07:29] so I don't really think thats true [07:29] cuz i do that quite often, that stacking you speak of dont work like you think,i'll time that too [07:29] s/work/as fast as [07:35] hello [07:35] hi stevanr [07:35] need some help.. [07:35] hi bkerensa [07:35] ello [07:36] after bootstrap I get https://pastebin.linaro.org/558/ [07:36] imbrandon: so that initial branch creation time was about 21s, vs. pushing a single small commit, which was 11s. :-P I am defeated. Please lets try to make bzr's death painless. ;) [07:36] heh mine is STILL pushing [07:36] here's my conf file https://pastebin.linaro.org/557/ [07:36] so umm no [07:36] imbrandon: still not 3m ;) [07:36] imbrandon: but what are you pushing?! [07:36] omg [07:36] imbrandon: mine spent 10s finding what to stack on, then the other 10s comparing graphs [07:36] imbrandon: so you're not stacked [07:37] bholtsclaw@ares:~/Projects/local/charms/precise/omg-bzr$ time bzr push lp:~imbrandon/+junk/omg-bzr [07:37] Warning: Permanently added 'bazaar.launchpad.net' (RSA) to the list of known hosts. [07:37] Created new branch. [07:37] real 4m19.228s [07:37] user 0m1.091s [07:37] no but wait, now i'm gonna stack one [07:37] sys 0m0.306s [07:37] imbrandon: yeah +junk is *ALWAYS* going to suck [07:37] *always* [07:37] ok ok , they all do for me [07:37] seriously [07:37] all the same, i never get fast bzr pushes [07:37] stevanr: reading [07:38] SpamapS, tnx :) [07:38] stevanr: can you use a differen't pastebin? [07:38] stevanr: my phone is downstairs.. don't feel like getting it for the 2-factor auth [07:38] SpamapS, lol, sry [07:38] i never understood why pastebin has a sso anyhow [07:39] wth is up with that [07:39] imbrandon: fast is relative. Its not going to beat git. We know that. But 10s to push 1 rev up is ok. Not great, but ok. [07:39] imbrandon: no, he did linaro pastebin [07:39] ahh [07:39] SpamapS: yea but no joke i mean the diff between minutes and seconds, not 11s and 22s, i could live with that [07:39] stevanr: just use 'pastebinit' .. should end up on paste.ubuntu.com [07:39] pastebinit for the win although I wish it supported more pastebins then just Ubuntu Paste [07:39] SpamapS: if he has it installed otherwise "sudo apt-get install pastebinit" [07:40] imbrandon: so, you're using something called +junk .. and you expect it to be .. not sucky? [07:40] SpamapS: lenaro and ubuntu both use sso [07:40] SpamapS: thats the first time ever i've pushed to junk [07:40] imbrandon: I have to agree that 3m is *WAAAAY* too long for new giant branch pushing. [07:40] SpamapS, ok so here's the status message after bootstrap http://paste.ubuntu.com/1014345/ [07:40] seeing 2kB/s at any time when I'm on my 30down/5up connection makes me think the other side of the pipe is on holiday [07:41] SpamapS, and here's my environments.yaml file http://paste.ubuntu.com/1014347/ [07:41] seriously some evening we'll get a byobu session going over here and i'll show ya , but thats some other day, but for real man i'm not exagerating at all on this one [07:41] stevanr: your machine agent is broken [07:41] agent-state: not-started [07:42] stevanr: with local provider, that starts at 'bootstrap' [07:42] SpamapS, i know :) [07:42] SpamapS, my question is 'why' [07:42] stevanr: it keeps a log, I forget where.. checking now [07:42] /var/logs/juju-machine* [07:42] iirc [07:43] stevanr: ahh, it would be in /tmp/local-dev/youruser-local/machine-agent.log [07:43] imbrandon: not on local [07:43] local provider just cocks everything up :-P [07:43] oh [07:43] its amazing when it works [07:43] but *EVERYTHING* seems different [07:43] yea i would think making it the same would be a goal [07:43] SpamapS, cool tnx, i'll get back soon [07:44] imbrandon: apt-get update/apt-get upgrade to get juju-jitsu 0.10 .. tested and works now [07:44] nice, kk [07:45] # jitsu run-as-hook mysql/3 config-get [07:45] {u'tuning-level': u'safest', u'query-cache-size': -1, u'query-cache-type': u'ON', u'binlog-format': u'MIXED', u'max-connections': -1, u'preferred-storage-engine': u'InnoDB', u'dataset-size': u'1G'} [07:45] unnnhh [07:45] thats hot [07:46] stevanr: that file should have clues. There's also an upstart job running it, so make sure it is not in the wrong state (like stop/waiting) [07:46] sweet if that works here thats EXACTLY what i wanted [07:46] i woudl kiss you but your married [07:47] doh [07:47] no key in the env on the nodes [07:48] but thats ok, i can add it to the install hook to set it [07:48] imbrandon: what key? [07:49] jitsu run-as-hook omg-wp/3 config-get [07:49] Could not find AWS_ACCESS_KEY_ID [07:49] imbrandon: you don't run it on your local machine [07:49] thats was form the node [07:49] actually ON omg-wp/3 [07:49] imbrandon: hm, I don't get any such errors [07:49] Unpacking replacement juju-jitsu ... [07:49] Setting up juju-jitsu (0.10-0stable1~precise1) ... [07:49] (juju-jitsu) root@ip-10-226-55-144:~# jitsu run-as-hook omg-wp/3 config-get [07:49] Could not find AWS_ACCESS_KEY_ID [07:49] (juju-jitsu) root@ip-10-226-55-144:~# [07:50] imbrandon: very weird [07:50] env , see if its set from something else [07:50] imbrandon: running on ec2 as well.. no such error [07:50] and no such env var [07:50] let me log out and back in, i was already in a jitsu wrapper [07:50] when i updated [07:50] imbrandon: I wonder.. are the access-key/secret-key set in the environments.yaml there? [07:51] oh yea, i have a env.y on the node [07:51] why ? [07:51] thats probably confusing it [07:51] its a bit of a hack [07:51] (juju-jitsu) root@ip-10-226-55-144:~# ls -l .juju/environments.yaml [07:51] -rw------- 1 root root 211 May 30 06:12 .juju/environments.yaml [07:51] (juju-jitsu) root@ip-10-226-55-144:~# [07:52] yup , moved it out of the way and works perfect [07:52] sweet [07:52] imbrandon: sounds like a bug tho [07:52] is that how its telling if its local or remote ? [07:53] jimbaker: ^^ note that run-as-hook doesn't work right when a local ~/.juju/environments.yaml is present [07:53] imbrandon: no I suspect its doing something wicked w/ the juju internal libs [07:53] OK I ran destroy-environment and bootstrap again and now it's in "running" state [07:53] seems like it could just look for /var/lib/cloud/instance/* exists instead of env.y [07:54] oh yeah, its importaing all kinds of weird stuff from juju's python libs [07:54] stevanr: awesome [07:54] yea [07:54] stevanr: would love to stay and help but I need to hit the sack [07:54] SpamapS, ok :) [07:54] SpamapS, tnx [07:55] SpamapS, your awakes.. [07:55] run as hook is crack [07:55] don't use it [07:55] yeah [07:55] reading it now [07:55] please remove it [07:56] hazmat: it seems to work tho ;) [07:56] SpamapS, it was to easy to make a miskate [07:56] hazmat: you have some way we can run hooks outside unit agent execution? [07:56] if you mis-understand it [07:56] and blow up your local machine [07:56] well. maybe overly dramatic.. but cause issue [07:56] SpamapS: hahah thats exactly whats its doing [07:57] def get_zk_client_connector(options): [07:57] 36 + # Running on a Juju machine or on an admin system? Distinguish by [07:57] 37 + # first checking for an ~/.juju/environments.yaml file; if not [07:57] 38 + # check the machine for an upstart conf file [07:57] hazmat: sounds like a reason to be wary, not to remove it [07:57] 39 + env_config = EnvironmentsConfig() [07:57] 40 + try: [07:57] 41 + env_config.load() [07:57] 42 + except juju.errors.FileNotFound: [07:57] hazmat: I would like a real way to do what it enables tho [07:57] SpamapS, sounds like a reason to enforce wariness... [07:57] if you have another way i'm all ears :) [07:58] i hate layout landmines for users [07:58] er. laying out [07:58] hazmat: jitsu isn't about wariness.. its about crack actually. Fresh, strong crack. [07:58] SpamapS, it really shouldn't be [07:58] hazmat: so, your open-port is totally legit? :) [07:58] Spamap, it is [07:58] until we change ZK [07:58] then its crack too [07:58] SpamapS, no, then its just broken [07:59] SpamapS, it won't damage your system accidentally [07:59] SpamapS, its not modifying zk directly, its using the api [07:59] you really haven't explained what run-as-hook does dangerously [07:59] well i could get this info from zk with phpzk, thats where its gonna end up anyhow is in php [07:59] if you like that idea better L( [07:59] :) [08:00] hazmat: I believe you, but I want something that does exactly this for a number of reasons. [08:00] SpamapS, executing hooks locally, completely changes any notion of env for a hook. [08:00] I don't know what you mean by locally [08:00] client side [08:00] I run it on the unit I want the info from [08:00] it works at that level [08:00] this isnt running client side we;re on the unit [08:00] can't we just enforce that? [08:01] like, you can only connect to the socket of the unit you're pretending to be? [08:01] why enforce things like that ... [08:01] wasted energy [08:01] imbrandon, because say you start a backup or delete files in such script [08:01] intending it for remote use [08:01] you point it to someone on irc [08:01] and they execute it locally by accident [08:02] ... [08:02] landmine [08:02] with great power comes great responsibility , we;re not talking about desktop users here, we;re talking what should be competent devops [08:02] then we should stop recommending jitsu to anyone not in charmers [08:02] what landmine has me handing out scripts that people blindly run? [08:03] SpamapS, they carefully examine it, and think it runs remotely [08:03] on the unit [08:03] thats why we changed it to run-as-hook [08:03] Its pretty obvious to me that this is just giving you a hook context locally [08:03] changing the camo on a landmine, doesn't change the nature of the landmine [08:04] what's the underlying common use case? [08:04] I can't imagine anyone ever going "Oh this will some how magically be transported to the unit and run there" [08:04] inspect/get the relation data [08:04] hazmat: all i'm saying is we cant protect them from everything esp when we;re making a sword, there is bound to be some shard parts [08:04] SpamapS, i think its documented that juju is 'cloud magic' ;-) [08:04] hazmat: but anyway, we can fix that really easily [08:05] that could hurt ppl, just like the dd tool installed on every desktop could wipe a hdd [08:05] hazmat: lets just require that the socket for the unit be accessible [08:05] SpamapS, +1 on the latter [08:06] bah that will just make those that dont want to jump that hook do it a real crackfull way like editing zk or mongo [08:06] document it , say here be dragons and let them cut a finger off once or twice [08:07] imbrandon, didn't you just use it for the time 10m ago? [08:07] i can promis that dd is installed on ever ubuntu system and can do just as much dammage [08:07] imbrandon, and are you executing it from the client? [08:08] hazmat: no i'm not ut my plan was not to be on the node 100% of the time either [08:08] bug 1006277 opened [08:08] <_mup_> Bug #1006277: run-as-hook needs to require the target unit agent to be local < https://launchpad.net/bugs/1006277 > [08:08] imbrandon: we can make a --force ;) [08:09] ok [08:09] that works [08:09] the thing has crap online help anyway [08:09] usage: run-as-hook [-h] [-e ENVIRONMENT] [--loglevel CRITICAL|ERROR|WARNING|INFO|DEBUG] [--verbose] SERVICE_UNIT COMMAND [ARG [ARG ...]] [08:09] expose a service unit(s) port [08:09] doh [08:09] heh [08:09] ok, sreep now kurinto san tired [08:09] gnight [08:09] imbrandon, its because the 'machine' identity is never really established.. sure dd can screw up a disk, but A) you know what machine its running on, ie. zero ambiguity, B) you need sudo ;-) [08:10] gnight too hazmat if your headed off [08:10] with run-as-unit.. A) is never really clear [08:10] imbrandon, just woke up.. have a good one [08:10] SpamapS, g'night [08:10] SpamapS: fixed the maintainer meta data thanks for the heads up [08:10] gnight SpamapS [08:10] ahh i'm not going , was just thinking all was on the same timeline a;) [08:11] hazmat: yea i see what your saying but i also see this as a "professionals tool" like a skill saw or similar [08:11] dangerious yes, you better know how to use it [08:11] etc [08:11] imbrandon, right.. but in that case its very different then the other jitsu commands [08:11] hazmat: as in juju as a whole not just jitsu [08:13] yea, i dont see the harm is making it very hard, as long as i can have --let-me-do-the-crack_sudo-make-me-a-smammich true [08:13] :) [08:13] s/is/in [08:14] i still dont see where it could wipe things local though [08:14] * imbrandon looks some more [08:15] hazmat: and really i think the GOOD(tm) awnser to this is for me and juan to finish the REST API. or learn go quicker so we can do it there w/e [08:16] cuz thats all i really want, is the info that may only be available durring a hook execution but the API would store and expose [08:16] ya know ? [08:18] bkerensa: the system nice ? desktop or laptop ? semi afk now while i play with my newfound crack [08:19] imbrandon, i'd really like feedback on the rest spec, its something that needs to be in juju, and i'm happy to shape jrapi into that, but the spec itself needs to get proposed first or it becomes a port issue [08:19] oh, i did not know the spec was actually down someehwre [08:20] i;ll start looking at it today then, and yea i'm more than happy to help aslo shape it into that and test it as well [08:20] is it a spec/blueprint on LP ? [08:20] or in the docs ? or elsewhere ? [08:21] hazmat: the spec is not that diff. from jrapi ... the changes wouldn't be that deep [08:21] and yea i totaly agree about the porting thing [08:21] negronjl: heya [08:22] imbrandon, hazmat: the README in jrapi is quite close to the spec [08:23] sweet, thats a good thing right hehe [08:23] negronjl, still needs fairly large changes.. testing, rewiring the existing client commands to use the rest api, deploying the rest api as part of bootstrap, upload support, auth, etc... fwiw.. i'm planning on starting on the REST work mid june. [08:23] probably mostly implmentation details [08:24] heh well hopefully i dont bug you toooo much as i likly wont be able to wait that long to begin [08:25] but yea whatever IS done between now and then should be in line with the goals ( looks like it already is started that way ) so its not tottaly way off [08:26] when the time comes [08:27] who knows, when i get to the office tomarrow they may tell me i got 6 months of nose to the keyboard stuff to do and i wont get any of this done [08:27] heh [08:27] * imbrandon knocks on wood [08:29] im out for the night ... [08:29] night man [08:29] take it easy [08:31] imbrandon: laptop [08:31] imbrandon: http://i.imgur.com/8L7hf.jpg === NCommand1r is now known as NCommander [09:12] <_mup_> juju/trunk r539 committed by kapil.thangavelu@canonical.com [09:12] <_mup_> [merge] show-errors-when-acquiring-node, use human error consumable error [09:12] <_mup_> messages when no maas nodes are available. [a=julian-edwards][r=fwereade,hazmat][f=980855] [09:12] hazmat, ty; sorry, I didn't see the changes go in :( [09:13] fwereade, no worries [09:14] fwereade, i haven't merged the port one [09:14] fwereade, i think it needs to handle https port 443 better.. ie introspect protocol to get default port [09:14] instead of just blindly using 80 if not specified [09:15] hazmat, hmm, good point === 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 [13:46] Hi! I am not sure if Rackspace is already supported by Juju? If not will it be in future? [13:48] doitdist_: Rackspace will be supported soon. Currently juju supports OpenStack but only with the EC2 and S3 middleware API turned on [13:49] ok thanks. what time scale is "soon" :-) [13:49] within a year? or 6 months? or no idea [13:50] I can't say for sure since I'm not a juju dev, but there's a preliminary branch with basic support. I imagine by 12.10 release - but that's just a guess [13:50] ok thanks! [13:50] np! [13:50] It's a pretty hot topic, given there are so many public OpenStack installs starting up, including Rackspaces [13:52] yes right... [13:54] how far are rackspace from being a normal openstack service these days? [13:54] mgz: public beta [13:55] ace. because their legacy bits don't work with core things like cloud-init. [13:55] mgz: I know, I hear their goal is to go live by the end of the year with OpenStack === mbarnett` is now known as mbarnett [14:57] pos/win 21 [15:53] 'morning all [15:58] 'morning all [15:58] * negronjl is reviewing the queue today [15:59] negronjl: saweeet [16:00] I just checked the calendar and I'm not in it to review but, there is nobody to do it either.... If I'm stepping on anyone's toes by reviewing, let me know ... [16:05] bkerensa: 01:10 < bkerensa> SpamapS: fixed the maintainer meta data thanks for the heads up [16:05] bkerensa: subway still has no maintainer [16:06] negronjl: go for it man. the pilot thing is just a commitment.. you can review wheneve ryou want to :) [16:06] SpamapS: thx ... Just making sure I don't step on anyone's toes :) [16:20] SpamapS: why? [16:20] SpamapS: I added the info to the metadata.yaml [16:21] whether it has been merged yet is another question [16:21] bkerensa: and where did you push that to? [16:21] bkerensa: I don't see anything in the review queue :) === hspencer is now known as hspencer[afk] [16:45] SpamapS: I will have to check [17:40] bkerensa: my bad, its there [17:40] bkerensa: just so new it doesn't have 'days' next to it, so I missed it. ;) [17:42] bkerensa: merged [17:49] SpamapS: ahh ;) [17:51]