/srv/irclogs.ubuntu.com/2014/09/24/#juju.txt

mwenninglazyPower, you still out there?00:07
mwenningok, catch you in the morning00:08
marcoceppiaisrael: yup, cheetah, but do whatever you want00:10
marcoceppiyou like Jinja? use that00:10
mwenninggood night!00:24
aisraelmarcoceppi: ack. Either one will be new to me, but I'll take a poke around and see which one works the best for me.01:28
=== thumper is now known as thumper-cooking
=== uru_ is now known as urulama
=== thumper-cooking is now known as thumper
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
=== ashipika1 is now known as ashipika
gnuoyI'm trying to do juju sync-tools but I'm getting this error http://paste.ubuntu.com/8417850/ . Does that look like a corrupt entry in the tools already uploaded ? And if so how do I see what tools the maas environment already has ?12:46
gnuoyjuju help sync-tools12:49
gnuoymgz, sorry to ping you directly but do you know where maas stores its copy of juju-tools ?12:54
marcoceppignuoy: doesn't it go in the blobstore in postgresql that maas exposes?13:11
gnuoymarcoceppi, that sounds like a distinct possibility, I'll have a look at the database13:12
mgzgnuoy: that is indeed the case, you can also use the maas cli to inspect13:45
gnuoymgz, excellent. I think I'm hitting a bug tbh but the maas cli should help me confirm13:47
gnuoymgz, do you know what the maas cli command is ?13:48
gnuoymgz, I see files list but no sign of tools http://paste.ubuntu.com/8418226/13:50
lazyPowersebas5384: i just saw you accepted the invite. o/ how many will be attending?13:57
sebas5384hey lazyPower, don't know yet13:58
lazyPowerOkie dokie13:58
sebas5384but I think at least two people13:58
sebas5384i'm going to the hangout :)13:58
sebas5384lazyPower: https://plus.google.com/hangouts/_/canonical.com/drupal-aumlet here?14:02
lazyPoweryep14:02
lazyPoweri'm already there14:02
sebas5384thats wired14:03
sebas5384ohhh hangout trolling our life again14:03
lazyPowerhaving an issue joining?14:03
=== kadams54_ is now known as kadams54
gnuoynatefinch, sorry to bother you again but does this error http://paste.ubuntu.com/8418336/ with sync'ing tools ring any bells with you ?14:04
natefinchgnuoy: no bother, it's specifically my job to help ;)14:05
gnuoy\o/14:05
natefinchgnuoy: this part looks suspicious: invalid binary version "1.21-alpha1.1--amd64"14:06
gnuoyagreed14:06
natefinchlike, why --?14:06
gnuoynatefinch, my guess is a bug in trying to decode the version data embedded in the tools filename14:06
natefinchahh, yeah, I think I get it... it's not --   it's -<series>-  ... except series is empty for some reason14:07
natefinchso, something is failing to parse the series from win2012r214:08
natefinchor, it's failing in the process somewhere, since we're not seeing an actual error until it looks for the binary14:09
jamespagegnuoy, I think we have some of our joined hook conditional processing based on being leader wrong14:10
natefinchgnuoy: sorry, in a meeting right now, but I can help later today.  in #juju-dev is gsamafira who wrote a lot of the windows support code, and he might be able to help, too.  I pinged him there, but seems like he's AFK right now.14:15
natefinchgnuoy: what timezone are you in?14:15
gnuoyuk14:15
gnuoyutc+114:15
natefinchgnuoy: ok, I'm us-east, UTC-4 so, will try to get back before your EOD14:19
natefinchahh, there's gsamfira.  gnuoy is having a problem deploying a windows charm: http://paste.ubuntu.com/8418336/14:19
natefinchnoteably: agent-state-info: invalid binary version "1.21-alpha1.1--amd64"    where -- is -<series>-   ... so series is getting set to an empty string at some point14:20
gsamfiragnuoy: if you look on the state machine, do you see any errors in the logs14:21
gsamfirasomething like "could not find tools for"14:21
gnuoyotp will look in a sec14:21
gsamfiraalso, how are you uploading the tools?14:22
gsamfiranvm14:22
gsamfirasync-tools14:22
gnuoygsamfira, with sync-tools (I included the sync-tools command in the pastebin)14:31
gsamfiragnuoy: can you try the latest master of juju. You seam to be using an older version  (it should be alpha2). Also, please provide the logs of your state machine. Also, please use hyper-v server 2012 R2 instead of windows server 2012 R2. You do not need windows server just for nova-hyperv. Hyper-V server 2012 installs quicker, and does not require a license.14:31
gnuoygsamfira, I think the statemachine may contain the smoking gun14:31
gnuoygsamfira, http://paste.ubuntu.com/8418508/14:32
stokachumarcoceppi, the test runner for charms does that basically juju deploy a charm in lxc and run whatever is in tests/*.test?14:33
gsamfiragnuoy: looks like the tools finder might be the problem. I have not touched that in a long time, but someone else might have. Are you using the latest trunk?14:35
stokachumarcoceppi, also does manage.jujucharms.com interpret TAP output?14:35
gsamfiralet me do a quick test with the latest master14:36
rick_h_stokachu: no, it doesn't.14:38
stokachurick_h_, how does the testing show the report? just pass/fail if the test returns error code?14:38
gnuoygsamfira, so I need to. Upload a Hyper-V server 2012 image to maas and switch to using juju 1.21-alpha2 . I built jujud.exe on a windows server 2012 R2 server can I still use that binary or do I need to rebuild with the hyperv edition (I'm guessing I don't need to) ?14:39
rick_h_stokachu: yes, it used to be that manage would check the test results in jenkins. With the new test suite I don't believe they're hooked up at the moment.14:39
gsamfirayou can use the same binary14:39
gsamfiragnuoy ^14:39
gnuoygsamfira, thanks14:39
rick_h_stokachu: it's something that has to be worked out as the testing tools get closer to general consumption14:39
rick_h_stokachu: so the TAP note is interesting14:39
gsamfirathe thing is, that windows server 2012 R2 does not have the hyper-v role enabled by default AFAIK14:40
gsamfiragnuoy are you using fastpath for windows? :)14:40
rick_h_stokachu: but would have to work with through jenkins and the test runner it looks like. Not sure if it's on their radar at atm.14:40
stokachurick_h_, yea, i think having TAP would be beneficial as people could write tests in whatever language they want14:40
stokachurick_h_, so the test runner does jenkins deploy the charm and then run the tests?14:40
rick_h_stokachu: I believe so. It handles the bootstrap/etc and then then the tests run against the deployment.14:41
stokachurick_h_, ok cool, thanks for the response14:41
rick_h_stokachu: np, thanks for the TAP earworm, something to think about.14:42
stokachurick_h_, np, there isn't a place to file enhancement requests for the manage.jujucharms.com app is there?14:42
rick_h_stokachu: well, it's not really manage. It's really just the consumer. I'm looking to see if the testing stuff is up somewhere a bug can be filed. tvansteenburgh have the link handy?14:43
stokachuah ok14:43
* tvansteenburgh reads scrollback14:44
rick_h_tvansteenburgh: is there a place to file bugs against the charm testing work?14:44
* rick_h_ is looking through GH and LP for something that looks like 'charm testing'14:44
gnuoygsamfira does editing the node in maas to set the os and release  set the installer correctly or is there another step there ?14:44
gsamfirayou only set the OS in MaaS if you plan on deploying it via MaaS. Juju takes care of requesting the proper OS from MaaS14:45
gsamfiraso no need to worry about that14:45
gsamfiragnuoy: the only thing to worry about when deploying windows server 2012, is that you have the KMS keys set up so it won't prompt for a serial key when installing14:46
gnuoykk14:46
gnuoygsamfira, but thats not a problem when using the hyperv image, right ?14:46
gsamfiragnuoy: right14:47
gnuoyexcellent14:47
tvansteenburghrick_h_, stokachu: https://github.com/tvansteenburgh/charmguardian14:47
tvansteenburghthat's what actually kicks off a test in jenkins14:47
gsamfiraHyper-v server does not require a key. Its free to use, unlimited, bla bla bla14:47
tvansteenburghrick_h_, stokachu: feel free to file stuff there, i'll move it if necessary14:48
rick_h_tvansteenburgh: ty much14:48
tvansteenburghthat repo will move under juju-solutions GH namespace soon, just haven't gotten around to it yet14:49
stokachutvansteenburgh, cool, so what causes a test to fail the return code of each test?14:49
tvansteenburghstokachu: correct14:49
stokachutvansteenburgh, and tests are run like ./tests/00-basic.test individually?14:50
stokachuor does it run like nosetests or something14:50
tvansteenburghstokachu: tests can be anything executable in charmdir/tests/14:50
stokachuok cool14:50
tvansteenburghit will also run `make lint` and `make test` targets if they exist14:50
tvansteenburghstokachu:  ^14:50
gsamfiragnuoy: http://paste.ubuntu.com/8418625/ <-- latest trunk14:51
stokachutvansteenburgh, cool thats good to know, the lint tool could it be a custom one?14:51
tvansteenburghstokachu: yeah, whatever you want14:51
stokachuthats pimp14:51
gnuoygsamfira, fantastic, let me give that a try14:52
stokachuso the frontend just parses this json  output then?14:52
tvansteenburghstokachu: yep, that's right http://reports.vapour.ws/charm-tests-by-charm14:52
stokachutvansteenburgh, and also what about lint output ?14:52
tvansteenburghstokachu: all output is captured and included in the test report details14:53
stokachutvansteenburgh, ok cool so long as my output is readable and comprehensable that would be ok?14:53
tvansteenburghstokachu: e.g. lint output: http://reports.vapour.ws/charm-tests/charm-bundle-test-942-results/charm/charm-testing-hp/114:54
tvansteenburghstokachu: yep14:54
gsamfiragnuoy: if you like, you can get the binaries from: http://gaby.rohost.com/tools/14:54
stokachusweet14:54
stokachutvansteenburgh, so with this setup would it even make sense to support TAP output?14:54
tvansteenburghstokachu: sorry i don't know what TAP is?14:55
stokachutvansteenburgh, http://testanything.org/14:55
* tvansteenburgh looks14:55
stokachutvansteenburgh, i use it a lot with my cpan modules14:56
stokachubut it works across several languages14:56
tvansteenburghstokachu: interesting. patches welcome :)14:58
gsamfiragnuoy: or even set your tools-metadata-url: to that value in environments.yaml14:58
stokachutvansteenburgh, cool, i can work on that if you aren't opposed to it14:58
tvansteenburghstokachu: i'm not opposed, i'm just fairly certain that it won't be something i have time for in the near future14:59
gnuoygsamfira,  I'll give the binary I built a try but if that fails I'll definitely grab those, thanks14:59
stokachutvansteenburgh, no worries i can take a stab at it and open a PR14:59
gsamfiragnuoy: my pleasure14:59
sebas5384thanks! lazyPower++15:05
sebas5384:)15:05
lazyPowersebas5384: my pleasure. Looking forward to the follow up where we can sink our teeth into your use cases15:06
sebas5384lazyPower: sounds great!15:06
lazyPowersebas5384: i'll schedule out a 2 hour block next time as well. Get me the list of names so we can get everyone together at teh same time isntead of a staggered start.15:06
sebas5384yeah definitively, and now that the team get all excited about, we are going to have more people contributing into it15:08
sebas5384:)15:08
sebas5384lazyPower: when you can send us the link of the recorder hangout, there are people that want to see already hehe15:10
lazyPowersebas5384: sure, let me export it, and edit off the extra 5 minutes i just recorded15:10
* lazyPower forgot i was recording15:10
sebas5384hehe15:11
lazyPowersebas5384: exporting now and i'll upload it unlisted to youtube for your team. incoming link shortly.15:13
sebas5384lazyPower: thanks :)15:14
whitnice definition of "Da Cloud": https://wiki.mozilla.org/CloudServices#Get_Involved15:18
=== arosales_ is now known as arosales
=== fabrice is now known as fabrice|family
=== bladernr_ is now known as bladernr_30kFeet
aisraelI think a charmer needs to look at this and change the status of this MP (disapproved): https://code.launchpad.net/~jaywink/charms/trusty/postgresql/swiftwal_missing_functionality/+merge/23539416:19
=== urulama is now known as urulama-afk
themonkJoshStrobl, hi16:50
themonki am facing a strange problem16:50
themonki set relation in join hook in charm 116:51
themonkand try to get that data in charm 2 in change hook, but getting None type value16:52
themonkmy other charm are working ok16:53
themonkcan any one give me any hint16:54
stokachutvansteenburgh, is make lint run in the deployed charm or just in the checked out repo directory16:55
stokachuor should i just have a target to make sure required dependencies are installed for a custom lint program to run16:57
tvansteenburghstokachu: the latter16:58
stokachutvansteenburgh, ok cool, thanks again16:58
tvansteenburghnp16:58
tvansteenburghcd /tmp16:59
* JoshStrobl was afk17:02
=== ev__ is now known as ev
=== JoshStrobl is now known as JoshStrobl[AFK]
=== mup_ is now known as mup
=== roadmr is now known as roadmr_afk
=== jog_ is now known as jog
rcjCan I set constraints in environments.yaml?  Specifically instance-type for EC2.19:15
=== CyberJacob|Away is now known as CyberJacob
Tugrcj: You don't set instance-type in evironment config19:39
TugHere is the list of available params: http://juju-docs.readthedocs.org/en/latest/provider-configuration-ec2.html19:40
Tugyou can set machine constraints when you deploy a service19:40
Tughttps://juju.ubuntu.com/docs/reference-constraints.html19:40
TugSome of my instances are scheduled for reboot by aws. I'm wondering is it something juju can handle ?19:49
TugLike did you think about restarting all the units on reboot?19:49
rcjTug, I want to avoid 'juju bootstrap --constraints="instance-type=m3.medium"' by having an setting in environment.yaml if it were available.19:50
Tugrcj, yeah but as you see, it has been deprecated19:51
Tugnot sure it's working today19:51
lazyPowerTug: once your api controller fires back up and your agents re-connect it will fire a config-changed event across the nodes19:52
Tugok thx lazyPower, so I guess config-changed should actually start the process if it's not running19:54
Tugbut it's the job of the charmer to know that19:54
lazyPoweryeah, i've seen a lot of the charms fire off hooks/start19:54
lazyPowerwhich is a decent pattern to adopt. It'll be a telling story as to which services dont come back online -  also - to note, if they have good upstart scripts you should be fine19:55
Tugmm not all charms run the underlying service using upstart.19:56
TugFor instance mongodb is run as a shell command from a python script with a deamonize option19:56
Tuganyway it's good to know19:58
lazyPowerTug: are you running the MongoDB charm in production?19:59
TuglazyPower, yes19:59
TugI have actually modified it19:59
lazyPowerwe should talk - as i'm the present hot seat maintainer for the MongoDB charm - anything that you found that was papercut worthy is totally worth investigating20:00
lazyPoweri'm still hacking through other priorities but its high on my list of TODO's to triage it in its current state and give it more love/tests20:00
Tugsure :)20:00
TugI haven't hacked on it for a while20:00
Tugbut I can try to help20:01
lazyPowerare you running a single node, or cluster?20:01
lazyPowerand is it replicated/sharded if cluster?20:01
TugI was running a 12 instances cluster (3 shards)20:01
TugI'm actually removing 2 shards at the moment20:01
Tugbecause it's starting to be quite expensive ^^20:02
lazyPowerindeed. Have you looked at running them on DigitalOcean?20:03
lazyPowerDO makes for a zippy MongoDB host, as their VPS's are SSD backed, and a fraction of AWS costs20:03
TugNop. I have free credits on amazon at the moment (startup credits)20:03
lazyPowerah, very nice20:03
sarnoldfor the micro instances?20:03
sarnoldthose things are slooooow :)20:04
Tugmongod are running on m3.medium instances20:04
Tugwith ebs provisioned iops storage20:05
Tugsarnold, nop it's real credits for everything on aws for 2 years20:06
Tuglimited to a few K20:06
=== roadmr_afk is now known as roadmr
Tuganyway, this was my branch https://code.launchpad.net/~dekervit/charms/precise/mongodb/trunk20:07
sarnoldTug: oh cool :)20:08
lazyPowerahhh i remember this branch20:08
lazyPowerand the comment on the bug before about the replicaset relation-changed method needing refactoring20:08
Tugyes, it was me :)20:09
Tuglike 3 months ago20:09
Tug;)20:09
lazyPowermhmm20:09
lazyPowerprogress has been slow, i dont have the time to really manage teh charm like it should be managed20:09
TugI haven't followed juju development for a while now. I guess you added ways to do automated testing and stuff20:10
natefinchrcj: it would be worth writing up a bug asking to support constraints in the environments.yaml.... that seems like a pretty nice feature20:10
rcjnatefinch, sounds good20:11
lazyPowerTug: Yep. Amulet testing, bundletester, etc. I'm working on the charm authorship docs as we speak actually20:11
rcjnatefinch, looks default instance type was removed20:11
TuglazyPower, nice, I'll want to have a look :)20:12
lazyPowerhttp://github.com/juju/docs - if you want to star/watch the repo you can get up to the minute updates20:13
natefinchrcj: yeah, I think that's been gone ever since we switched away from pyjuju .... the instance-type constraint was only added back fairly recently (April).  It should be pretty trivial to add support for reading constraints from environments.yaml, I'd think.20:13
TuglazyPower, done20:13
rcjnatefinch, cool20:14
Tuganyway I might get back on mongodb charms soon. If I can make it stable enough maybe it can be merged into the main branch ;)20:14
lazyPowermbruzek, marcoceppi - https://github.com/juju/docs/pull/17220:19
natefinchrick_h_: let me know if I screwed anything up in that doc ;)20:25
rick_h_natefinch: thanks looking20:26
mbruzeklazyPower:  To quote tvansteenburgh, "I saw what you did there, and I didn't like it"20:28
mbruzeklazyPower: There is some merge text in your document you want to clean up.20:28
lazyPowerwhat?20:28
lazyPowerlooking20:29
mbruzeklazyPower: your pull request20:29
* lazyPower throws a fit20:29
lazyPowerthats it, i'm going back to vim diff20:29
lazyPowerscrew meld20:29
lazyPoweroh wait, it has the .orig file included20:29
lazyPowerwhat kind of nonsense is this, thats user error20:29
lazyPowermbruzek: corrected and repushed20:31
lamontis there really no named charm?20:33
lazyPowerlamont: there's a start to a dns-charm that would have support for named20:33
lazyPowerbut its nowhere near ready for public consumption20:33
lazyPowerlamont: https://github.com/chuckbutler/DNS-Charm20:34
lazyPowerthere are spec docs included in the charm if you want to lend a hand20:34
lamontlazyPower: I'll look.20:34
lamontI also have an openvpn-server charm to upload, that actually charms the things I want charmed (no, the CA does not belong anywhere near the openvpn endpoint)20:35
lazyPowerlamont: stuff it in your namespace, and open a review ticket against it :)20:36
lamontcould you expand upon "my namespace"?20:37
lazyPowerlamont: https://juju.ubuntu.com/docs/authors-charm-store.html#name-space-charms20:37
lamontthank yo20:37
lazyPowerlamont: the idea is its 2 stage. No reviews required for your charm to be listed in the charm store under your namespace. When you're ready to undergo the review process to become a recommended charm - you then progress to the next stage. This is 2 fold 1) this enables you to get moving quickly and have your charm listed in the charm store without high walls to hurdle. 2) you can get community feedback on your charm through use of your charm in your20:39
lazyPowernamespace, before you have a charmer look at it  (but we're open to reviewing charms anytime regardless of recommended status desire - so feel free to open a review ticket at any time)20:39
lazyPowerhazmat, kwmonroe, mbruzek, marcoceppi - https://github.com/juju/docs/pull/181  - Digital Ocean provider docs22:24
kwmonroelazyPower: the v1 api stumped me for a bit earlier this week.  glad you got explicit with the screenshots.  i'll circle back tonight for a merge if nobody has other comments.22:33
lazyPowerack. Thanks kwmonroe22:33
arosaleslazyPower: DO doc commented on23:28
arosalesthats https://github.com/juju/docs/pull/181 for referernce.23:28

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