=== Beret- is now known as Beret | ||
=== ming is now known as Guest12249 | ||
=== menn0-afk is now known as menn0 | ||
=== jcw4 is now known as jcw4_zzz | ||
AskUbuntu | juju server relation hook files are not running? | http://askubuntu.com/q/505310 | 06:01 |
---|---|---|
=== CyberJacob|Away is now known as CyberJacob | ||
=== CyberJacob is now known as CyberJacob|Away | ||
AskUbuntu | juju charm relation-joined hook not working? | http://askubuntu.com/q/505342 | 08:27 |
=== CyberJacob|Away is now known as CyberJacob | ||
jamespage | marcoceppi, hey - do you think it would be possible to have an openstack-charmers review queue like we have a charmers on on jujucharms.com? | 09:09 |
jamespage | marcoceppi, I'm struggling with visibility of proposed changes right now and a central report would be useful for everyone | 09:10 |
jamespage | gnuoy, ^^ | 09:10 |
=== CyberJacob is now known as CyberJacob|Away | ||
=== vila is now known as vila-lunch | ||
=== CyberJacob|Away is now known as CyberJacob | ||
=== CyberJacob is now known as CyberJacob|Away | ||
marcoceppi | jamespage: yes, I'm in the process of re-doing the review queue to be way more robust. In doing so it'll be a stand alone application that anyone can run and configure to track changes for whatever user/group/project | 11:24 |
marcoceppi | as such, we could spin up an instance on canonistack to track openstack-charmers stuff | 11:24 |
=== vila-lunch is now known as vila | ||
jamespage | marcoceppi, sounds good | 11:57 |
marcoceppi | jamespage: it's a little lower priority than everything else, but it's on my personal "I really care about this and want it done" list | 11:58 |
jamespage | marcoceppi, if you have something in flight maybe post a branch? I'm happy to hack on this as well | 11:58 |
marcoceppi | jamespage: I have early musing of some pyramid stuff, nothing connected to lp or gh yet | 11:59 |
marcoceppi | just a databases schema | 11:59 |
marcoceppi | jamespage: https://github.com/marcoceppi/review-queue | 11:59 |
marcoceppi | I hope to get more time on it this weekend, get celery and lp hooked up for initial imports | 12:00 |
mfa298 | I've got a Juju instance bootstrapped within Openstack using trusty. Is it possibly to use that to deploy a charm that uses precise or is that not supported (I've got the precise image in glance but can't see a way to tell Juju where to find that image). | 12:48 |
marcoceppi | mfa298: you'll need to upload a custom image-metadata file to juju | 12:51 |
marcoceppi | so it'll know where the precise images are | 12:51 |
mfa298 | that's presumably created with juju metadata generate-image, how do I then upload it ? | 12:52 |
marcoceppi | mfa298: great question, I forget how, but I believe it's done at bootstrap time | 12:52 |
* marcoceppi checks | 12:52 | |
mfa298 | so it may not be possible to upload after bootstrap | 12:53 |
marcoceppi | mfa298: does not appear, but again I'm not 100% certain, it's something that's defined in the environments.yaml https://juju.ubuntu.com/docs/config-openstack.html | 12:54 |
marcoceppi | However, let me check set-environment | 12:54 |
marcoceppi | mfa298: you can update this after bootstrap | 12:55 |
marcoceppi | with `juju set-environment image-metadata-url="url-to-generated-metadata"` | 12:55 |
mfa298 | hmmm, looking at the metadata that generate-metadata created for the precise image it seems to reference 14.04 rather than 12.04 | 13:00 |
marcoceppi | mfa298: you should be able, as a command line option, provide a series flag | 13:02 |
marcoceppi | mfa298: with the -s flag | 13:02 |
mfa298 | I'd just spotted that | 13:02 |
mfa298 | looks like juju still can't find the image | 13:08 |
marcoceppi | mfa298: what do the logs look like when trying to deploy a precise charm? | 13:09 |
mfa298 | debug-log shows machine-0: 2014-08-01 13:07:20 WARNING juju.worker.instanceupdater updater.go:231 cannot get instance info for instance "": no instances found | 13:10 |
mfa298 | and the machine state in juju status shows: agent-state-info: '(error: index file has no data for cloud {RegionOne http://192.168.17.17:5000/v2.0} | 13:10 |
marcoceppi | mgz_: who should I bug about juju deploying to openstack? | 13:11 |
mfa298 | trying juju metadata validate-images seems to suggest it's using cloud-images.ubuntu.com rather than the local images, although that could be me missing something | 13:12 |
mgz_ | marcoceppi: that'd pretty generic | 13:12 |
mgz_ | what more specifically? | 13:12 |
mgz_ | I'm a pretty good atrting point | 13:13 |
marcoceppi | mgz_: mfa298 is trying to upload image-metadata post bootstrap, not sure the process as I've never tried | 13:13 |
mgz_ | I see, reading log | 13:13 |
mgz_ | setting up image-metadata is something you do prior to bootstrap | 13:18 |
mfa298 | so having bootsrapped juju with it only knowing about trusty there's not a way to add in precise other than destroying the environment and bootstrapping again ? | 13:19 |
mgz_ | yup | 13:19 |
mfa298 | or is this going down the wrong route for what I was hoping to acheive. | 13:20 |
mgz_ | if you're setting this up yourself, you should instead make your keystone advertise the simplestreams | 13:20 |
mgz_ | rather than have juju supply it at run time | 13:20 |
mfa298 | quickly googled and that looks like it might be what I want. Is there a decent guide somewhere for setting that up ? | 13:23 |
mgz_ | mfa298: https://juju.ubuntu.com/docs/howto-privatecloud.html | 13:26 |
marcoceppi | mgz_: whoa, where has this link been all my life | 13:27 |
marcoceppi | also, wow tha page is rendered wrong | 13:28 |
* marcoceppi goes to patch | 13:28 | |
mfa298 | mgz_: I've done the juju metadata generate-image commands and have the metadata files which contain both the precise and trusty files. | 13:35 |
mfa298 | the issue seems to be getting that into juju | 13:35 |
mfa298 | I can run juju set-environment image-metadata-url=file://home/ubuntu but I still don't seem to be able to deploy a charm using precise | 13:36 |
mfa298 | or is the answer here that I need to have a web server to provide those files rather than file:// | 13:37 |
mgz_ | no, that doesn't help | 13:37 |
mgz_ | you set those values in your environments.yaml at the start, and they need to be accessible from the cloud you've deployed, eg in swift | 13:38 |
mgz_ | file:// is no good | 13:38 |
mfa298 | so installing a webserver would be enough or do I also need to destroy the environment and bootstrap again with the url configured as well | 13:39 |
mfa298 | looks like a http server is enough. That seems to be working | 13:49 |
mfa298 | thanks | 13:49 |
mfa298 | now to work out the sets of commands that were actually needed so I can write the local documentation | 13:52 |
jcastro | sebas5384, arosales: hi | 13:56 |
sebas5384 | hey jcastro! | 13:57 |
arosales | sounds like jose has us set up | 13:57 |
jcastro | oh cool | 13:57 |
jcastro | link? | 13:57 |
arosales | just in case I am going to joing the old hangout in case any folks join there | 13:57 |
jose | ohai | 13:57 |
jose | jcastro: sec | 13:57 |
sebas5384 | link? | 14:00 |
sebas5384 | :P | 14:00 |
arosales | Just a quick reminder to folks that we will be hosting on ubuntuonair.com not via the google hangout event. | 14:01 |
* marcoceppi tunes in | 14:03 | |
arosales | I confirmed no folks are in the Google Event hangout, which is good | 14:03 |
jcastro | sebas5384, https://plus.google.com/hangouts/_/ytl/vjVQa1PpcFNNnOvmc6cVeIQVTxHO4-o-wB5b0uus7JY | 14:05 |
arosales | also we'll be using the following document to capture input | 14:06 |
arosales | https://docs.google.com/a/canonical.com/document/d/1t_55N1il3XoL8z-jfa1CBoSxzOQjC90cgSpCqx5wkH0/edit | 14:06 |
arosales | we'll be starting in a couple of minutes. | 14:06 |
lazyPower | ah ok we haven't started yet. I was following along on Ubuntu on Air and its still broadcasting please stand by. | 14:06 |
arosales | lazyPower: correct haven't staretd yet. | 14:07 |
jose | :P | 14:07 |
sebas5384 | i'm there! | 14:08 |
jose | we should have a slide saying 'Hey! We're late, but don't go!' | 14:08 |
arosales | jose: looks like we are running into a perms issue | 14:10 |
arosales | jose: are you in the hangout? | 14:11 |
jose | arosales: I am | 14:11 |
arosales | jose: jorge and I are in the hangout but don't see you. | 14:11 |
jcastro | jose, can you paste the link in here? | 14:11 |
jcastro | the one you PMed me isn't the one you are in apparently | 14:11 |
cory_fu | Regarding the first point, I think what we really need is an extension to `juju resolved --retry` that essentially does a forced upgrade-charm before retrying the failed (or maybe even last successful) hook. `juju resolved --upgrade-and-retry` | 14:11 |
sebas5384 | i'm already there, but i cant hear jose | 14:11 |
jose | https://plus.google.com/hangouts/_/ytl/vjVQa1PpcFNNnOvmc6cVeIQVTxHO4-o-wB5b0uus7JY | 14:11 |
zirpu | i thought the hangout wasn't being used today. | 14:12 |
jose | I could go ahead and create another event | 14:12 |
jose | arosales, jcastro: want me to create another event? | 14:13 |
jcastro | Why don't we all just join the same hangout? | 14:13 |
jcastro | I am confused why there are two? | 14:13 |
sebas5384 | jose: can you hear me? | 14:13 |
jose | sebas5384: not at all | 14:13 |
sebas5384 | damn it | 14:14 |
sebas5384 | hangout is trolling us | 14:14 |
jose | blame Google | 14:14 |
jose | aaaalways happens | 14:14 |
jcastro | what hangout are you in? can you paste in the URL? | 14:14 |
jose | jcastro: https://plus.google.com/hangouts/_/ytl/vjVQa1PpcFNNnOvmc6cVeIQVTxHO4-o-wB5b0uus7JY | 14:14 |
sebas5384 | https://plus.google.com/hangouts/_/g6mlkq4hfo6jvgjqmvksxo3inia?authuser=1&hl=pt-BR | 14:14 |
jose | wait... that's another hangout link | 14:14 |
jose | urgh | 14:14 |
jcastro | ... | 14:14 |
jose | I believe Google is playing with us | 14:14 |
jcastro | ok which hangout link are we going to? | 14:15 |
jose | we have different links | 14:15 |
jcastro | pick one | 14:15 |
jose | https://plus.google.com/hangouts/_/ytl/vjVQa1PpcFNNnOvmc6cVeIQVTxHO4-o-wB5b0uus7JY is the one I'm in | 14:15 |
sebas5384 | permissions problems | 14:15 |
jose | let me just quickly create another event. Google is a mess atm | 14:16 |
jcastro | do you have them set to private or something? | 14:16 |
jose | not at all | 14:16 |
jcastro | hmmm | 14:17 |
jcastro | ok fire up a new one I guess | 14:17 |
jose | https://plus.google.com/hangouts/_/ytl/vjVQa1PpcFO4tsipIQdBcOcv_43jywHwKQvrzxaa9XA= | 14:17 |
zirpu | the ubuntu-on-air one was supposed to replace the hangout. | 14:17 |
jose | jcastro, arosales, sebas5384: ^ | 14:17 |
sebas5384 | now i'm in the last link you passed jose | 14:18 |
arosales | this one is looking better | 14:18 |
jose | zirpu: ubuntuonair uses hangouts :) | 14:18 |
arosales | jcastro: I am in | 14:18 |
jose | refresh ubuntuonair if you were in before | 14:19 |
arosales | we got it started | 14:20 |
arosales | Thanks for the patience. | 14:20 |
arosales | live on ubuntuonair.com or join the hangout @ https://plus.google.com/hangouts/_/ytl/vjVQa1PpcFO4tsipIQdBcOcv_43jywHwKQvrzxaa9XA= | 14:21 |
zirpu | i switched to the ubuntuonair version. i'm just listening. | 14:22 |
cory_fu | You can give debug-hooks a hook name to only have it trigger on the one hook you're interested in fixing. | 14:35 |
cory_fu | I just learned this the other day. Very helpful | 14:35 |
cory_fu | jcastro: If local-mapped-to-remote charm source is not an option, my preferred alternative would be `juju resolved --update-and-retry` | 14:37 |
lazyPower | thats good to know cory_fu. We should make that more pronounced in the docs, which might help alleviate some of that frustration. | 14:38 |
lazyPower | noodles775: you've done roles with ansible scripts in charms correct? | 14:46 |
lazyPower | i'm fairly sure its still very experimental at present right? we haven't ironed out how it shoud look | 14:47 |
lazyPower | jcastro: we dont have --force on service. | 14:47 |
jcastro | ack | 14:48 |
arosales | lazyPower: thanks. | 14:48 |
cory_fu | If you --force destroy the machine out from under the service, you can then remove the service | 14:48 |
zirpu | i think the ansible buffering is a function of the parallelism code. so make ansible serial and lower the polling from the default 15 seconds. | 14:48 |
lazyPower | cory_fu: i think the idea is more that they want to force destroy the service, and leave the machine, so they aren't waiting for a machine spin up | 14:49 |
lazyPower | jcastro: you're left with a service definition, with no units. | 14:49 |
cory_fu | lazyPower: If you redeploy after removing a service, it creates a new machine. If you use --to to put it on the same machine, that's not much different than upgrade-charm. But I guess it would force it to re-run all of the hooks, at least. | 14:50 |
lazyPower | well, the scary part about what's being asked is deploying to a tainted machine may yield really crazy results | 14:51 |
lazyPower | but i get what you're saying cory_fu | 14:51 |
* cory_fu isn't arguing against adding --force to destroy-service, though. | 14:51 | |
cory_fu | I've tried to do that many times, even after realizing it doesn't work. | 14:51 |
lazyPower | haha | 14:51 |
lazyPower | it gets me too | 14:51 |
lazyPower | more often than i care to admit... and i know it doesn't exist. | 14:51 |
lazyPower | muscle memory i suppose | 14:51 |
cory_fu | :) | 14:52 |
arosales | lazyPower: is this: http://manage.jujucharms.com/~lazypower/precise/dns your latest DNS charm? | 15:13 |
sebas5384 | arosales: Handrus and Renato where here too :) | 15:13 |
arosales | ah thanks Handrus and Renato! | 15:14 |
arosales | sebas5384: take a lookt at http://manage.jujucharms.com/~lazypower/precise/dns just need to confirm this is the last rev from lazyPower (re DNS) | 15:15 |
lazyPower | arosales: it is. i sync'd it a few weeks ago witht he latest work. | 15:15 |
arosales | lazyPower: thanks for confirming. | 15:16 |
jose | hey guys! question around here. I'm working on a chamilo-memcached relation, and it would allow multiple servers. when I do 'relation-get host', will it tell me just one IP address, or multiple IP addresses? | 15:21 |
jose | (in the event I have multiple memcached instances) | 15:22 |
benji | now | 15:23 |
benji | wrong window :) | 15:23 |
marcoceppi | jose: you can infer all the hosts at once if you wanted to | 15:25 |
marcoceppi | jose: using relation-list | 15:25 |
marcoceppi | jose: then just loop through the list | 15:25 |
jose | hmm, I'm gonna check how that may work for me in a debug-hooks session | 15:26 |
marcoceppi | memcached_hosts=`relation_list`; for m in "$memcached_hosts"; do relation-get host $m >> /file/to/track/hosts; done | 15:26 |
marcoceppi | as an example | 15:26 |
jose | cool | 15:26 |
marcoceppi | the relation-get is from memory | 15:26 |
marcoceppi | but there's a way to specify which unit you wish to query in a relation context | 15:27 |
=== jcw4_zzz is now known as jcw4 | ||
jcastro | lazyPower, https://github.com/juju/docs/pull/135 | 15:45 |
sebas5384 | lazyPower: thanks!!! could you show us how to use it? http://manage.jujucharms.com/~lazypower/precise/dns | 17:33 |
=== roadmr is now known as roadmr_afk | ||
=== StoneTable is now known as aisrael | ||
lazyPower | sebas5384: its not production ready yet. There's no HA support as of yet. | 19:23 |
lazyPower | if you use that, and your DNS charm server tanks, you've lost DNS | 19:23 |
lazyPower | sebas5384: but i'm more than happy to talk you through the implementation details, and how its structured / how to implement hooks. | 19:23 |
lazyPower | sebas5384: take a look at https://github.com/chuckbutler/DNS-Charm - and scroll down to CHARM Integration, it talks about a programmable and programmable-multiple relationship hook. You set the proper variables, and it will build the configuration on the fly for you. The DNS charm itself spits out the public-address OTW so you can update /etc/resolve.conf as the primary DNS server, and your domains will then be avialble to each node connected to the dns | 19:25 |
lazyPower | charm. | 19:25 |
lazyPower | sebas5384: there's more implementation logic that needs to happen with regard to updating third party providers, and/or implementing your DNS server in the global DNS tree with your registrar (if you want it to be authoritative) | 19:25 |
jcastro | lazyPower, hey so btw my last PR didn't touch the precise box URLs | 19:30 |
jcastro | so you might want to pull those | 19:30 |
lazyPower | ack was already on it | 19:30 |
lazyPower | jcastro: just in master? or do i need to touch another branch? | 19:31 |
lazyPower | jcastro: https://github.com/juju/docs/pull/136 | 19:34 |
=== rektide_ is now known as rektide | ||
jcastro | lazyPower, LGTM, merged | 19:49 |
jcastro | lazyPower, marco told me a while back it's better to just do all the work in personal branches and then submit to master | 19:50 |
jcastro | rather than under the juju namespace | 19:50 |
jcastro | I was like, ok, sounds good to me | 19:50 |
lazyPower | jcastro: thats what i did | 19:50 |
jcastro | yeah, I saw | 19:50 |
jcastro | I was just responding to your irc question | 19:50 |
lazyPower | oh you mean the web editor | 19:50 |
lazyPower | well i was curious which branch to target | 19:50 |
lazyPower | if i needed to touch the 1.18 docs as well | 19:50 |
lazyPower | not that we are still actively pointing anything at them | 19:50 |
lazyPower | actually no need to sync, just riddle me this batman | 20:07 |
lazyPower | https://code.launchpad.net/~asanjar/charms/trusty/hdp-hadoop/trunk <-- hortonworks? | 20:07 |
lazyPower | https://code.launchpad.net/~asanjar/charms/trusty/hdp-zookeeper/trunk <-- has first traces of hortonworks charm helpers? | 20:08 |
asanjar | lazyPower: yes, that is hortonworks .. but if you need to investigate bdutils.py (general big data charm helper) or hdputils.py (hortonworks dirstro specific charm helpers) look at ~asanjar/charms/trusty/hdp-zookeeper | 20:13 |
asanjar | lazyPower: as soon as I get a chance, I will update hortonworks hadoop with the latest helper.. | 20:15 |
lazyPower | ok thats all i needed | 20:16 |
lazyPower | I'm wrapping up my last fringe issues thsi week with vagrant that jcastro just brought to me | 20:16 |
lazyPower | Monday is when i start digging heavy into teh apache hadoop rewrite | 20:16 |
=== roadmr_afk is now known as roadmr | ||
=== CyberJacob|Away is now known as CyberJacob | ||
=== CyberJacob is now known as CyberJacob|Away | ||
=== CyberJacob|Away is now known as CyberJacob | ||
AskUbuntu | Swift Through Horizon | http://askubuntu.com/q/505650 | 23:31 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!