/srv/irclogs.ubuntu.com/2014/03/14/#juju.txt

* Makyo is away: EoD00:17
bodie_oh that's annoying00:36
bodie_https://github.com/kapilt/juju-digitalocean requires manual deployment of VMs before you can put services on them00:36
bodie_doesn't that kinda defeat the purpose?00:36
davecheneybodie_: that is why it's called the 'manual' provider :)00:37
bodie_*mutters*00:38
rick_h_bodie_: heh, yea that's the advantage of using full providers00:39
davecheneycall it a forcing function00:39
bodie_digitalocean just got $37M in funding.. maybe we should think about enhancing juju's interface with 'em at some point ;) they're trying to take on AWS00:40
bodie_really hits the price/performance sweet spot with all the extra fluff removed00:41
bodie_not that EC2 and S3 are fluff... but for a lot of people, the bells and whistles just complicate things00:42
bodie_just thinking out loud here :)00:42
rick_h_I'm sure others have pondered and sent an email or two :)00:43
=== whit_ is now known as whit
=== Kyle- is now known as Eliz
davecheneybodie_: the ball is in their court, they need to provide an API00:56
davecheneyand join the CPC program00:56
joselazyPower: hey! around?03:48
marcoceppijose: I think he's off for the evening03:48
josemarcoceppi: oh, I had a question (as he was the last to review the mailman charm)03:49
joseI'm about to do a push and wanted to see if someone could test it (local provider has some problems as you know)03:49
marcoceppijose: I can try03:50
josesure, I should push it in ~3m03:50
josemarcoceppi: mind trying the charm? config-changed was giving an error before, should be fixed now03:53
marcoceppijose: linky link?03:53
josehttps://bugs.launchpad.net/charms/+bug/1199052 and https://code.launchpad.net/~jose/charms/precise/mailman/trunk03:53
_mup_Bug #1199052: New charm: mailman <Juju Charms Collection:New> <https://launchpad.net/bugs/1199052>03:53
joselp:~jose/charms/precise/mailman/trunk03:53
josemarcoceppi: any blockers found?04:09
marcoceppijose: http://paste.ubuntu.com/7088242/04:09
josemarcoceppi: you know a way of using spot instances with juju? like, manually? I *may* be able to afford a couple to test04:10
marcoceppijose: the manual provider can do that04:11
marcoceppijose: what's wrong with local provider?04:11
joselast time I tried using there was a bug on LXC which prevented me from deploying04:11
joseI'm going to try again04:11
joseERROR cannot find network interface "lxcbr0": net: no such interface // ERROR failure setting config: cannot find address of network-bridge: "lxcbr0" // ERROR cannot find address of network-bridge: "lxcbr0"04:12
josemarcoceppi: ^04:12
marcoceppijose: sudo apt-get purge lxc juju-local04:13
marcoceppisudo apt-get install juju-local04:13
marcoceppisee if that creates the lxcbro04:13
* jose does04:13
davecheneyubuntu@winton-02:~/src/launchpad.net/juju-core$ juju deploy cs:trusty/ubuntu04:15
davecheneyERROR charm not found: cs:trusty/ubuntu04:15
davecheneyY NO TRUSTY CHARM!!!04:15
marcoceppibecause we no promulgate trusty charms yet04:16
davecheneyY U KEEP CHRM SECRETT04:17
josemarcoceppi: oh, do you have a link to those trusty tests docs you mentioned?04:17
marcoceppidavecheney: because charms have no tests!04:18
marcoceppijose: which trusty tests docs?04:18
joseyou mentioned having some docs about writing the tests for charms to be on trusty04:18
davecheneymarcoceppi: ubuntu charm is awesome, it is the chuck norris of charms04:18
davecheneyit doesn't need tests04:19
marcoceppidavecheney: you bring up a good point though, we need to start testing trusty like tomorrow04:19
davecheneymarcoceppi: i need to test it like months ago :)04:19
marcoceppidavecheney: I can push the ubuntu charm to trusty if you'd like04:20
marcoceppiit's obviously going to work04:20
marcoceppidavecheney: cs:trusty/ubuntu exists now, may take a few mins for the store to find it04:23
marcoceppidavecheney: nope, it found it https://store.juju.ubuntu.com/charm-info?charms=cs:trusty/ubuntu04:23
josemarcoceppi: juju first downloads the precise image before deploying, right?04:24
marcoceppijose: yes, the precise clodu iamge04:24
josethen it'll take a good while04:24
joseI better go grab a cup of coffee04:25
davecheneymarcoceppi: w00000000t04:29
davecheneythanks04:29
davecheneyi can use a local charm04:30
davecheneybut the whole test is that we can talk to thet charm store via a proxy04:30
davecheneyso local charms defeate the purpose04:30
marcoceppidavecheney: I want to enable you FOR GREATNESSS04:30
davecheneythank you sir04:30
davecheneydo you have a suitable imgur for this ?04:30
marcoceppihttp://i.imgur.com/1ida17K.jpg04:34
marcoceppinot relaly04:34
sarnoldawesome though :)04:34
marcoceppithough I've been abusing this one recently, davecheney, http://i.imgur.com/9EqJvVX.gif04:34
josemarcoceppi: hey, it says 'instance-state: missing'04:42
marcoceppijose: that's progress, what does ~/.juju/local/log/machine-0.log have for contents?04:43
josehmm, /me checks04:43
josemarcoceppi: http://paste.ubuntu.com/7088329/04:44
marcoceppijose: it looks like it's still progressing04:45
marcoceppigive it a few more mins checking juju status occasionally04:45
josesure, thanks04:45
marcoceppijose: what does sudo lxc-ls --fancy show?04:46
josestate: running, gives me an ip address04:46
josemarcoceppi: mind a PM on another topic?04:47
marcoceppisure04:47
josemarcoceppi: I cannot seem to find the error on config-changed... Everything looks sane to me04:56
marcoceppijose: let me get you the log04:57
marcoceppihttp://paste.ubuntu.com/7088378/ jose04:59
bodie_night all05:00
joselet's see05:03
joseoh, I'm going to check that05:12
=== zchander__ is now known as zchander
evis there any way to stop juju from tearing down the instance when it fails to bootstrap, so I can investigate why?07:05
evI'm seeing:07:06
evBootstrapping Juju machine agent07:06
ev2014-03-14 06:51:09 ERROR juju.provider.common bootstrap.go:123 bootstrap failed: rc: 107:06
stubI don't suppose there is a daily package built of HEAD somewhere?08:06
stubhttps://code.launchpad.net/~dave-cheney/+recipe/juju-core-daily no longer seems to exist08:09
=== erkules_ is now known as erkules
JoshStroblHey guys, where is the Charm Bundle lesson?09:14
JoshStroblmarcoceppi: says you'll be on of the speakers, where's the stream?09:17
JoshStroblhmm, maybe my Google Calendar just got messed up or something and is displaying the wrong time to me09:18
JoshStrobloh yea, 1400 EST on Friday (just looked at the email). Good job Google, waking me up for no reason :\09:20
* JoshStrobl goes back to bed09:21
rbasakhazmat: got an issue with python-websocket-client, which python-jujuclient uses.12:20
rbasakhazmat: I need to shift the dependency to python-websocket, since we have a dupe. This involves effectively bumping the version in Trusty from 0.11.0-0ubuntu1  to 0.12.0-112:21
rbasakhazmat: do you see any problems with that?12:21
rbasakjamespage: ^^12:21
jamespagerbasak, going to give it a quick test as well12:21
jamespageI use deployer alot12:21
hazmatrbasak, should be fine.. i'll double check re changes12:24
hazmatrbasak, yup that's fine.. i use that version already12:24
hazmatvia pip/virtualenv installs12:25
rbasakhazmat: thanks!12:26
=== BradCrittenden is now known as bac
jcastrombruzek, lazyPower, those errors you got with the subordinates and the bundles last night, you got those in the GUI right?13:16
mbruzekjcastro, I was using the command line.  I did not see any errors, I only saw 3 charms deployed from that bundle.13:17
mbruzekI am sure there were errors that prevented that from deploying, but I didn't dig into it.13:18
rick_h_jcastro: the gui was watching the bundle and returned the error to the user13:18
rick_h_jcastro: but the error came from lower down the stack13:18
jcastrombruzek, can you see if you have the bundles you tried  in your history? I'd like to dig today13:19
mbruzekjcastro, http://pastebin.ubuntu.com/7090108/13:21
mbruzekI destroyed the environment at the EOD but that is how I ran it.13:21
jcastroah got it13:21
jcastroI mistakingly put in the "complex", but chuck caught that and I removed it13:21
jcastropushed a new version, trying it now13:21
mbruzekOut of that list of charms, I only saw juju-gui, kibana, and haproxy deployed iirc13:22
jcastroyeah, rails takes like 10 minutes at least13:22
jcastroit does all this ... rails stuff ...13:22
jcastropull the universe from git, etc.13:22
jcastroI am trying it now13:22
mbruzekOK13:23
zchanderHi all... Got another (weird) error. Just reinstalled my MaaS and whenI want to bootstrap my Juju, I get iscsi errors13:28
* zchander seems baffled.... Somehow some ephemerals were installed twice...... :(13:50
lazyPowerjcastro: i've tried it on several installations. the units were never coming online13:51
lazyPowerthat error failed the entire bundle deployment. i'm not saying its not my env being the culprit though13:51
jcastroyeah rick is investigating13:51
jcastrono I reproduced, and so did mbruzek just now13:51
lazyPowerah ok13:51
jcastrobut this exact bundle was working a few days ago, so clearly something changed13:51
jcastrolazyPower, I am worrying about the sample app in the rails charm though13:52
mbruzekjcastro, that pastebin was from yesterday13:52
jcastrombruzek, yeah the exact same thing happened to me13:52
jcastrojust now13:52
mbruzekOk13:52
lazyPowerjcastro: ok let me wrap this subordinate I am hacking on and I'll pull that sample rails app and see how it goes13:55
jcastroI am +1 if you want to replace pavel's app with another hello-world style app13:55
lazyPoweractually13:56
lazyPowerlet me deploy pavels app on the rails charm now. it wont take long13:56
rick_h_jcastro: [E 140314 09:52:59 utils:72] error deploying the bundle                                                                                                        │·······················13:56
rick_h_[E 140314 09:52:59 utils:73] error type: <class 'jujuclient.EnvError'>                                                                                         │·······················13:56
rick_h_[E 140314 09:52:59 utils:79] error message: subordinate service must be deployed without units13:56
jcastrohow did it let me create that in the first place?13:57
rick_h_jcastro: wow, this error was mentioned in our irc back in July13:58
jcastrowell, I prefer known to unknown. :)13:59
rick_h_http://irclogs.ubuntu.com/2013/07/24/%23juju-gui.txt line 7313:59
jcastrook so worse case, remove the subs14:00
jcastrobut then the bundle is kind of less useful14:00
rick_h_jcastro: pulling num_units and going to try it out again14:00
rick_h_jcastro: I think just removing the unit specifier will work, but testing14:00
jcastrook14:01
* Makyo is back (gone 13:44:06)14:01
jcastrorick_h_, the django bundle has the same errrors14:03
jcastroso hopefully that's the badger14:03
rick_h_jcastro: in looking at the source code here, num_units isn't valid param for subordinates. Did you export this? I thought we didn't export that value any more14:03
jcastroI exported it14:03
jcastroI've done no manual mangling other than renaming the envExport14:03
rick_h_jcastro: will file a bug on that then. We shouldn't export that setting14:04
jcastrook so I should remove that from the bundle and try again?14:04
jcastrorick_h_, remove all of the num_units right?14:04
rick_h_hmm, same error14:05
rick_h_jcastro: yea it's what I did to try out but failed again14:05
* jcastro stands by14:05
rick_h_jcastro: which are sub's in here?14:07
jcastrologstash agent14:08
rick_h_nrpe?14:08
jcastromaybe nrpe?14:08
jcastroI don't remember, I made the bundle like 3 weeks ago, heh14:08
jcastroiirc I thought subs would be mentioned in the file14:08
lazyPowerrick_h_: confirmed nrpe and logstash-agent are the subs14:09
rick_h_lazyPower: k, trying another take14:10
rick_h_jcastro: ok, change the num_units to 014:11
rick_h_on the subs14:11
rick_h_jcastro: don't ask me why, but it's what the code says and it's deployed the rest of the services. So run with me here :)14:12
rick_h_lazyPower: ^14:12
rick_h_for the django one14:12
rick_h_as well I'd imagine14:12
jcastroneither nrpe or logstash-agent have num_units mentioned14:12
jcastrodo you think it's assuming one? so explicitly add it and declare 0?14:13
lazyPowerrick_h_: http://i.imgur.com/00Ugaft.gif14:13
rick_h_jcastro: yes, it seems to assume 1 and then error. I set it to 0 and it's working14:13
jcastroman, that's an awesome bug14:14
rick_h_jcastro: pretty much14:14
rick_h_jcastro: k, all done. rails failed on install hook error but the rest went out14:16
rick_h_the sample app one that is14:16
marcoceppijcastro: I take issue with the way the rails bundle is structured14:17
jcastromarcoceppi, ok14:17
marcoceppiSince it's a "framework" charm it's still to just deploy it14:17
jcastrofire away14:17
marcoceppiWe should instead model some rails software in the bundle instead14:18
marcoceppilike rails/<rails-app-thin>14:18
marcoceppisince bundles should model workloads14:18
marcoceppifind popular rails apps and build bundles for them in the rails name space14:18
lazyPowermarcoceppi: I suggested stringer as a possible replacement.14:18
lazyPoweranother one would be spree, the ecommerce solution.14:18
marcoceppianother issue with the charm in general, but it's not your fault jcastro, is a lot of times it needs post-deployment commands to work properly14:19
jcastroyes, I had to mention that14:19
marcoceppi(db:migrate, asset"precompiile, etc)14:19
marcoceppijcastro: http://stackoverflow.com/a/487407/19683214:19
jcastrobasically, django and rails need things like, a config.yaml for where the code lives, etc.14:19
lazyPowermarcoceppi: since we cleared the security groups lastnight I haven't been able to reach anything i've deployed on HPCloud - is there a stray setting aside from the .jenv i need to wipe?14:20
lazyPowerwait, i can reach it via ssh, but not port80 post being exposed14:21
* lazyPower wat's14:21
marcoceppilazyPower: did you have anything deployed before the wipe?14:21
lazyPoweri did :(14:21
marcoceppiyeah, they're inaccessible14:21
lazyPoweri had a bootstrap node chillin in there that I manually wiped from the console.14:21
marcoceppibest to destroy environment and move on14:22
lazyPoweri just rebootstrapped14:22
* lazyPower wipes and tries again14:22
jcastromarcoceppi, ok how about example-single and example-scalable?14:30
jcastroto make it explicit that it's a hello world for you to customize?14:30
marcoceppijcastro: such compromise14:30
marcoceppijcastro: much synergy14:30
jcastrorick_h_, blamo, worked first time14:31
jcastrook going to do what marco wants, push, and then we'll see about the example app14:31
jcastrogoing to fix/push django too14:31
marcoceppiso, probably needs to be on deployer than14:32
rick_h_jcastro: woot thanks for the help in debugging. I've got a card for us to find the right place to file the bug.14:32
marcoceppia bug*14:32
=== hatch__ is now known as hatch
rbasakhazmat: about python-jujuclient again. test_jujuclient.py is in bzr, but not in the PyPI distribution, so I can't easily run the test suite as part of a package build (MIR requirement).14:58
rbasakhazmat: I just dealt with exactly this problem in websocket-client by plucking out the test suite in a packaging patch.14:58
rbasakhazmat: please can you add it to setup.py?14:59
hazmatrbasak, sure15:01
hazmatrbasak, i hadn't realized this was going into main15:01
hazmatrbasak, i've got meetings for the next 2hrs but will come back to it today15:01
rbasakhazmat: np. Unless you do a PyPI release and I update packaging I'll have to do a distro patch anyway.15:02
rbasakhazmat: it's going into main because juju-quickstart needs it.15:02
rbasakhazmat: "Seriously Alpha. Works now, but API *will* change." alarms me a little for main inclusion.15:03
* rbasak takes a break to unwind his stack15:04
hazmatrbasak, yeah.. i should take that out, it was commentary from the first release against the api, but in truth it has been pretty stable to date15:06
rbasakhazmat: thanks. I will quote that for the MIR :)15:07
onreafainally, I could not run Discourse.org by juju :(15:47
onreaMy problem is repored here: http://discourse.ubuntu.com/t/how-to-install-juju-manually-cloudimg-tar-gz-where-should-it-be-extracted/154815:48
onreaif you can help, please help15:49
onrealazyPower: ^15:56
lazyPoweronrea: ack, hang on i'm going to tap one of the more knowledgeable people15:58
marcoceppionrea: put them in /var/cache/lxc/cloud-precise but don't extract them16:00
onrea:)16:00
onreaboth of them?16:00
marcoceppionrea: yes16:00
onreaw816:01
onreamarcoceppi: PLease check the instruction for any missing/wrong step:16:03
onrea1. sudo juju destroy-environment -y local16:03
onrea2. juju init && juju switch local16:03
onrea3. sudo chown -R emzi.emzi ~/.juju ~/.ssh16:03
=== BradCrittenden is now known as ba
=== ba is now known as bac
onrea4. juju bootstrap16:03
lazyPoweronrea:  once you've dstroyed you don't need to re-init16:04
onrea5. juju deploy --repository=$HOME/charms local:discourse16:04
onrea6. juju deploy postgresql16:04
onrea7. juju add-relation discourse postgresql:db-admin16:04
onrea8. juju expose discourse16:04
onrea9. juju status16:04
onrea10. watch juju status16:04
onrealazyPower: ok16:04
lazyPoweronrea: otherwise that set of instructions loks fine, i don't see anything crazy16:05
onreaOk, thanks @lazyPower, marcoceppi. I'm going to rock discourse with juju...16:06
=== Ursinha is now known as Ursinha-afk
=== Ursinha-afk is now known as Ursinha
jcastroERROR cannot use 37017 as state port, already in use16:33
jcastroanyone know how to fix this one with the local provider?16:33
jcastroI've blown away local.jenv and the environment/local directory16:34
marcoceppijcastro: sudo initctl list | grep juju16:40
jcastrook I see stuff16:42
jcastrodo I stop/restart it?16:42
rick_h_jcastro: yea, I just ps aux | grep mongo and kill the mongodb instance running16:43
marcoceppijcastro: sudo stop each of those16:43
marcoceppithen rm -f /etc/init/juju-*16:44
jcastro==> unit-jenkins-slave-0.log <==17:37
jcastro2014-03-14 17:37:43 ERROR juju runner.go:220 worker: exited "upgrader": cannot read tools metadata in tools directory: open /home/jorge/.juju/local/tools/1.17.4.1-prec17:37
jcastroise-amd64/downloaded-tools.txt: no such file or directory17:37
jcastroI've not run into this before17:38
lazyPowerjcastro: this is something that happened during the 1.17.4 update - if you ln -s the saucy tools to precie it goes awy17:39
lazyPowerand its only log spam - it doesn't appear to have any effect over deployments in my testing.17:39
lazyPowers/precie/precise17:40
jcastroI don't have saucy or precise tools17:40
lazyPowerare you running trusty?17:40
jcastroyeah17:40
lazyPowerprobably trusty-tools then17:40
lazyPowersame principal17:40
lazyPowerbut ymmv - i dont know if there was a specific difference between the series.17:40
jcastroah17:41
jcastroI don't seem to have /releases/juju-1.17.4.1-trusty-amd64.tgz17:41
jcastrocan I sync-tools post-bootstrap?17:41
lazyPowergood question17:41
* jcastro tries it17:42
lazyPoweri think so17:42
lazyPowerjuju sync-tools is referenced in this answer17:42
lazyPowerhttp://askubuntu.com/questions/285395/how-can-i-copy-juju-tools-for-use-in-my-deployment17:42
jcastroit didn't work, but the service is up17:43
jcastroso Good Enough17:43
* lazyPower nods17:44
lazyPowerthe log spam is annoying, i'll file a bug17:44
jcastrowell, the tools are also missing17:45
jcastrothat might be a problem, heh17:45
lazyPoweryikes!17:45
lazyPowersinzui: anything going on with streams/tools that you are aware of?17:45
sinzuiNothing yet.17:46
lazyPowerbrb17:46
sinzuilazyPower, It is stable right now. When will ask for an update to get 1.17.5 when I see the arm64 and ppc packages in ports17:47
lazyPowersinzui: ack, thank you for looking into this for us17:53
jcastroJuju Charm School on Bundles in 2 minutes!17:58
jcastrohttps://plus.google.com/hangouts/_/hoaevent/AP36tYdbEgmEEjOzBoN2_eYNDyWicF98hHYXAdD2H92SaNCpwqY1DA?authuser=0&hl=en17:58
jcastrothere's the URL17:58
jcastrohttp://youtu.be/eYpnQI6GZTA if you just want to follow along17:58
JoshStroblWill the Youtube stream automatically start playing the video once it's being broadcasted?18:01
marcoceppiJoshStrobl: yes, should start in a few mins18:01
JoshStroblcool18:01
marcoceppiJoshStrobl: rather, it should be running now18:01
marcoceppimay need to refresh18:01
JoshStroblyep!18:02
JoshStroblrefreshed and it's buffering18:02
JoshStroblwell, it's loading as in waiting...or something like that. no way it's buffering on my connection18:02
lazyPowerJoshStrobl: its live for me - still having issues?18:03
JoshStroblnot any more. just ended up loading firefox and it was fine18:04
JoshStroblgood job Chrome, good job...18:04
* lazyPower golf claps for chrome18:04
marcoceppiFeel free to ask questions anyone watching, and I'll relay them in to the video!18:05
JoshStroblaanndd this is why I use nano in the terminal :P18:11
=== Ursinha is now known as Ursinha-afk
=== Ursinha-afk is now known as Ursinha
ppetrakiany questions so far?18:16
bacjcastro: the ingest runs every 15 minutes, so worst case is you just miss and it takes two cycles, 30 minutes18:17
JoshStroblNo questions from me :(18:21
JoshStrobl* :)18:21
lazyPowerjcastro: you should be able to drag directly from the rails charm to that mongos instance in the middle.18:23
lazyPowerjcastro: crazy? that was every day in a prior life :)18:24
=== Ursinha is now known as Ursinha-afk
=== Ursinha-afk is now known as Ursinha
JoshStroblhey lazyPower, I changed my charm from Incomplete to Fix Committed again barring a response from you (regarding my response to you): https://bugs.launchpad.net/bugs/1278369 . Waiting a while and noticed there wasn't any reply on it when it hadn't been changed, so I went ahead and changed it.19:07
_mup_Bug #1278369: Charm Needed: metis <new-charm> <Juju Charms Collection:Fix Committed by truthfromlies> <https://launchpad.net/bugs/1278369>19:07
lazyPowerJoshStrobl: thanks! i reached out and didn't receive an email response - so without it being in teh queue and no 1:1 from you it was pending re-entry into the queue19:08
lazyPoweri should be able to get you the +1 today before I EOD19:08
JoshStrobllazyPower: Ah, thought that since ~charmers were subscribed that maybe it'd notify others, guess not.19:08
lazyPoweryeah - with me not being a charmer yet (emailing the list today to petiton for membership) - i don't get those notifications yet.19:09
JoshStrobllazyPower: I changed to Fix Committed since it was more relevant to my actual post (I had changed to Fix Committed prior to you setting it as Incomplete since like 99% of everything marcoceppi talked about was resolved). Not necessarily looking for the +1 (although that'd be nice), more like "hey, the stuff you said wasn't resolved actually was, here is a long post about it".19:10
JoshStroblIf you have any questions and feel it'd be more appropriate to discuss on IRC rather than mailing list, I'll obviously be around...probably for 5 or so more hours.19:14
lazyPowerJoshStrobl: ack. I'll ping you shortly19:19
lazyPowerwrapping up some expedited requests now19:19
joselazyPower: hey, if you're around and have a minute, looks like I fixed the mailman charm19:28
lazyPowerjose: awesome :) did you move the status of your ticket to fix committed?19:28
joselazyPower: well, to new, but moved it19:29
lazyPowerack. as long as its in the revq i'll get to it before my shift ends for teh week19:29
lazyPowerJoshStrobl: ok moving on to metis - incoming review19:36
JoshStrobllazyPower: Thanks bud.19:37
lazyPowerJoshStrobl: without cryptograhic verification of file signatures I cannot pass the charm, its a hard stopper. I respect your concern of continuous delivery though19:45
JoshStrobllazyPower: Any suggests?19:45
JoshStrobl*suggestions19:45
lazyPowerMy thought would be to use git itself as the delivery mechanism, since git validates the file signatures cryptographically19:45
lazyPoweryou can use tags in the same vein as you would be zip packages19:46
lazyPowerthen you just deliver from your github repository using the tag, and make tag the configuration option for deployment19:46
lazyPoweror hash, or whatever. This satisfies a default stable release and allows for upstream deployment as well so you're not updating your charm over and over to get the new releases in19:46
JoshStroblGonna be honest, didn't even think about using git. Feel free to mark as Incomplete barring fix, I'll implement using git as the delivery mechanism when I get the chance and will change it to Fix Committed when it's done.19:48
JoshStrobl*git for delivery19:48
lazyPowerJoshStrobl: if you want any help/pointers ping me and I'll hop on a hangout with you19:48
JoshStroblCool, thanks!19:48
lazyPowerand thank you for being patient during the review process, i know this has quite a bit of back and forth. I want to see this land in the charm store soon after seeing how much effort you've put into it19:48
JoshStrobllazyPower: Thanks for your kind words and the tip on using git (makes sense since the code is on GitHub anyways) =)19:49
* lazyPower hat tips19:57
lazyPowerjose: incoming mailman review19:57
joselazyPower: thanks!19:57
=== hatch__ is now known as hatch
lazyPowerjose: correct me if i'm wrong, but i'm seeing the same failure20:49
lazyPowerwas this corrected on the postfix charm?20:49
joselazyPower: hmm? let me double check20:49
lazyPoweras that may be why i'm not seeing the fix show up20:49
joseno, it was corrected on the mailman charm20:49
joselet me double check just in case20:50
lazyPowerack20:50
josehmm, no, it was corrected on the 7th revision20:50
josewhich is on LP atm20:50
joseI did a successful deploy with it on my local env yesterday20:51
lazyPowerI'm sitting on revision 520:51
lazyPowerand it just updated20:51
lazyPowerwat...20:51
lazyPoweri'm slippin20:51
lazyPowerok let me upgrade the charm and start over - ty for confirming20:51
jose:P np20:52
lazyPowergood news! it passed the config-changed hook blocker20:54
lazyPowermoving on20:54
=== nszceta_ is now known as nszceta
=== hatch__ is now known as hatch
lazyPowerok last call for anyone with last minute charm submissions before i EOD21:18
lazyPoweryou've got 15 minutes21:18
* lazyPower EOD's21:36
lazyPowerjose: ping me when you've got those changes and i'll re-review for ya later tonight and get your +1 on record21:37
joseawesome, thanks!21:37
lazyPowerJoshStrobl: not sure if you're still hacking away, but same goes for you.21:37
=== lazyPower changed the topic of #juju to: Welcome!! Docs: http://juju.ubuntu.com/docs || FAQ: http://goo.gl/MsNu4I || Review Queue: http://goo.gl/9yBZuv || Unanswered Questions: http://goo.gl/dNj8CP
dpb1marcoceppi: hi -- amulet is supposed to intercept calls to deploy the "current" charm, right?  We are seeing that isn't happening as we expect.  Known issue?21:45
marcoceppidpb1: how are you executing the test with charm test or directly?21:46
marcoceppiif directly you'll need to see this Ask Ubuntu question21:46
dpb1hmmm.21:47
dpb1charm test?21:47
dpb1we are using juju test21:47
marcoceppidpb1: that's one in the same21:47
dpb1reading in a deployer config file with 'branch: lp:~landscape-charm/trunk'21:48
marcoceppijuju test == charm test; are you launching juju test from the charm_dir?21:48
dpb1yes21:48
marcoceppidpb1: OH21:48
marcoceppithat's why21:48
dpb1:(21:48
marcoceppiit only intercepts on d.add()21:48
marcoceppidpb1: file a bug and I'll see what I can do21:48
dpb1OK21:49
dpb1marcoceppi: there are other problems with the config-based deployment that I haven't filed yet.  But this one is more tricky for me to work around.21:50
ahasenackmarcoceppi: what do you whink about this other idea,21:51
ahasenackmarcoceppi: when creating a deployment, like21:51
ahasenack    d = amulet.Deployment(sentries=False)21:51
ahasenackmarcoceppi: pass it an existing directory where I have already checked out all the branches I want deployer to use21:51
ahasenackit's like pre-creating precise/<charms>, and then calling deployer21:52
ahasenackit will see the charms are already there and not update them, nor check them out again21:52
ahasenackso, I would call it like21:53
ahasenackd = amulet.Deployment(sentries=False, charm_directory="/some/path")21:54
ahasenackit would use charm_directory instead of a temporary directory to check out all the charms21:54
fishy_what username and password can I do to get into a new local charm I just started?22:54
fishy_assume I would do a sudo lxc-console --name <foo>22:54
joselazyPower: should be ready for review, but needs the 'password' option set for the install hook to run successfully23:06
lazyPowerjose: ack - do a sniff for the presence of the password config option, if its not present, dont run the hook23:06
lazyPowerthat breaks idempotency23:06
josehuh?23:07
joseoh23:07
josethen the charm would not be useful at all, because the service wouldn't even install23:07
lazyPower$HTPASSWDPW=`config-get password`    if [ -z $HTPASSWDPW ]; # do stuff  fi23:07
lazyPowerim' talking about securing the mailman config interface, dont run the htpasswd generation stuff if options aren't populated23:07
josehmm, good idea, not have the htaccess if password is null23:07
joseyep23:07
lazyPower:)23:08
joselazyPower: aaaaand, pushed!23:13
lazyPowerjose: woo! updating and reviewing now23:15
lazyPowerhttp://www.youtube.com/watch?v=21OwTUEiGGM#t=23123:15
lazyPowerwatch that while you wait23:15
jose:P23:15
lazyPowerjose: really close - move that logic to the config-changed hook, otherwise its immutable and only runs on install23:25
joselazyPower: hmm, in what part?23:25
joselike, adding support to change the password?23:27
lazyPowerjose: the heredoc configuration where you're generating the htpasswd and setting the apache values for the htpasswd23:27
lazyPowerright :)23:27
lazyPowerimmutable configuration options will wind up tanking a review just as fast as a broken hook - because its not readily apparent to the user that its only configurable during install, and thats a pretty easy copy/paste job to move it.23:28
joselazyPower: all set now23:32
lazyPowerdeploying23:35
joselazyPower: thanks a lot for your help, owe you a couple beers once we meet at a conference23:37
lazyPowerjose: nbd :) I enjoy the candor though23:37
lazyPowerjose:         agent-state-info: 'hook failed: "install"'23:47
joselazyPower: does it give you any clues on the log?23:47
joseI'll try and deploy, but will take a while23:47
lazyPowerln: failed to create symbolic link `/etc/apache2/sites-enabled/mailman': File exists23:47
josehmm? /me double checks23:48
joselazyPower: weird, I don't see that on my logs23:49
lazyPowerthats on a fresh local provider environment23:51
lazyPowerthe install hook is only run once though - so if you're updating a unit, you wont see that23:51
joseI just destroyed my local environment and bootstrapped again23:51
joseit's deploying23:51

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