wallyworld | then you copy the generated files up the the public bucket | 00:00 |
---|---|---|
wallyworld | hatch: i just tried juju --version and it works for me, what version of juju do you think you have installed? | 00:00 |
lamont | error: flag provided but not defined: --version | 00:00 |
lamont | 1.12.0-0ubuntu1~ubuntu12.04.1~juju1 | 00:00 |
hatch | wallyworld: I JUST did it this morning, following the 'install guide' | 00:00 |
lamont | anyway, afk for a couple of hours | 00:00 |
hatch | trying to see what the typical user would see | 00:00 |
hatch | lamont: thanks for confirming :) | 00:01 |
wallyworld | lamont: ok, ping us in #juju-dev if you want any more input | 00:01 |
wallyworld | hatch: i'm running from source, but "juju --version" worked for me | 00:01 |
sarnold | hatch: the python version and the go version have different --version vs version behavior. that's almost a decent way to tell which one you have already.... :) | 00:02 |
sarnold | wallyworld: (oh, has that been fixed?) | 00:02 |
wallyworld | i didn't realise there was a difference | 00:02 |
wallyworld | i've only ever run go juju | 00:02 |
hatch | oh that's odd :) | 00:02 |
sarnold | ah :)) | 00:02 |
wallyworld | ian@wallyworld:~$ juju --version | 00:03 |
wallyworld | 1.15.0-raring-amd64 | 00:03 |
hatch | sarnold: thanks for clearing that up | 00:03 |
sarnold | wallyworld: neat! yay :) | 00:03 |
hatch | 1.12.0-precise-amd64 | 00:03 |
hatch | here | 00:03 |
hatch | on stable | 00:03 |
wallyworld | ah ok. there's been lots of fixes post 1.12 i think | 00:03 |
wallyworld | we are looking to release 1.14 next week | 00:03 |
hatch | well then....someone should update stable :P | 00:04 |
wallyworld | for now, there's a 1.13.3 | 00:04 |
* hatch stokes fire | 00:04 | |
wallyworld | hatch: agree, i think that will happen | 00:04 |
wallyworld | each even number release is considered stable | 00:04 |
hatch | but really though thanks - it looks like the go version doesn't handle the -- on 1.12.0 | 00:04 |
wallyworld | appears so, sorry | 00:04 |
hatch | but if it works fine on your version then I won't file a bug | 00:04 |
wallyworld | ok | 00:05 |
hatch | this happened when running through the Local Configuration setup docs | 00:06 |
hatch | it says `juju generate-config --show` which failed | 00:06 |
hatch | just fyi | 00:06 |
wallyworld | hmmm. seems the docs may need some love then | 00:07 |
hatch | I'm guessing that it was just done with a more recent version | 00:08 |
hatch | unless even yours doesn't do --show :) | 00:08 |
marcoceppi_ | lamont: still having issues with private cloud? | 00:45 |
wallyworld | marcoceppi_: i think he's afk for a bit, but i got his public bucket url sorted, now he needs to generate image metadata | 01:14 |
marcoceppi_ | wallyworld: cool | 01:15 |
AskUbuntu | Juju e MAAS get error in apt | http://askubuntu.com/q/344064 | 01:41 |
=== medberry is now known as med_ | ||
=== freeflying is now known as freeflying_away | ||
=== freeflying_away is now known as freeflying | ||
=== CyberJacob|Away is now known as CyberJacob | ||
kenn | Question about the bridge node. Due to budget constraints I can only run a single instance in the cloud. Are there any reasons why I wouldn't want to deploy my services to that first machine 0, or anything I should be aware of when doing that? I've done it locally so it's possible to do. | 06:39 |
=== rogpeppe1 is now known as rogpeppe | ||
AskUbuntu | Can I deploy juju on Eucalyptus | http://askubuntu.com/q/344137 | 07:37 |
fwereade_ | jcastro, marcoceppi_: if either of you are around, I'd appreciate advice re resolving doc conflicts caused (at least partly) by header/footer changes | 08:08 |
fwereade_ | jcastro, marcoceppi_: eg "take your tree, copy in $files from trunk, run tools/build.py, commit, merge trunk in" | 08:09 |
=== defunctzombie is now known as defunctzombie_zz | ||
=== freeflying is now known as freeflying_away | ||
=== freeflying_away is now known as freeflying | ||
AskUbuntu | help: bootstrap error | http://askubuntu.com/q/344168 | 09:35 |
=== freeflying is now known as freeflying_away | ||
=== freeflying_away is now known as freeflying | ||
=== kentb-out is now known as kentb | ||
jcastro | fwereade_: evilnickveitch is your guy there | 13:17 |
jcastro | fwereade_: I am pretty sure he strips all that out and regens the footer/header anyway | 13:18 |
fwereade_ | jcastro, thanks, I chatted to him | 13:18 |
jcastro | marcoceppi_: also ... review queu! | 14:09 |
marcoceppi_ | jcastro: yup :\ | 14:09 |
=== freeflying is now known as freeflying_away | ||
evilnickveitch | jcastro, fwereade_ I just posted to the list about the new, super-easy way of creating pages I just sorted out this morning... | 15:02 |
jcastro | k | 15:03 |
fwereade_ | evilnickveitch, cool, thanks | 15:13 |
bloodearnest | heya all - am having some problems with juju-deployer (0.2.3) hanging after deploying all services, but before adding relations | 15:20 |
bloodearnest | Ctrl-C'ing gives me a tb: https://pastebin.canonical.com/97357/ | 15:21 |
bloodearnest | same tb every time | 15:21 |
bloodearnest | had this happen on 2 different raring machines, on both lxc and openstack (canonistack) envs | 15:23 |
bloodearnest | any pointers to fix? | 15:23 |
jcastro | http://pad.ubuntu.com/7mf2jvKXNa | 15:31 |
jcastro | T minus 30 minutes until the Charm Call! | 15:31 |
jcastro | 10 minutes until the juju charm call! | 15:51 |
=== BradCrittenden is now known as bac | ||
mattyw | jcastro, is there a link to just watch the hangout? | 16:09 |
marcoceppi_ | mattyw: ubuntu-on-air.com | 16:10 |
marcoceppi_ | mattyw: http://ubuntuonair.com/ | 16:10 |
mattyw | marcoceppi_, of course, thanks | 16:11 |
=== marcoceppi_ is now known as marcoceppi | ||
Chor | hi there | 16:48 |
kurt_ | Can you guys tell me if its correct that a deployment fails when there is an existing configuration file for a charm? I'm seeing this when having destroyed keystone, then trying to redeploy it. Maybe an adjunct questions is when a service such as keystone is destroyed, should it clean up it's configuration files cleanly? | 17:15 |
marcoceppi | kurt_: if you destroy a service, then try to deploy it again, I believe you get an error about service already deployed | 17:24 |
marcoceppi | kurt_: is that the error you're getting? | 17:24 |
kurt_ | marcoceppi: no, the problem, specifically with keystone was that the /etc/keystone/keystone.conf is left behind | 17:25 |
kentb | does anyone have a *fairly current* and recommended set of steps for deploying openstack charms with maas & juju, especially with quantum-gateway. For whatever reason, with quantum-gateway in the mix, keystone authentication with quantum gets hosed. every. freaking. time. | 17:26 |
marcoceppi | kurt_: that's a different issue | 17:26 |
sarnold | early charms might expect the unit to be destroyed when the service is unconfigured / terminated / etc. I'm not surprised it didn't do a great job of cleaning up | 17:26 |
kurt_ | kentb: I've been working on this for some time. I hope to have something out in the near future | 17:27 |
kurt_ | there are several guides out there, but you have to be patient and figure it out | 17:28 |
kentb | kurt_: yeah, that's the part that's about burned up (patience). I'll also take whatever you have in the meantime. The quantum-gateway piece is the one that I just can't seem to crack. | 17:29 |
kurt_ | marcoceppi: this appears to be some conflict in removing 2013.1.2-0ubuntu2~cloud0 and it's need for /etc/keystone/keystone.conf | 17:30 |
kurt_ | kentb: yes, I've ran in to this too. you have to plan your deployment topology very carefully | 17:30 |
kurt_ | kentb: have you seen jamespage's excellent guide? https://wiki.ubuntu.com/ServerTeam/OpenStackHA | 17:31 |
=== CyberJacob is now known as CyberJacob|Away | ||
kentb | kurt_: yep. I've worked off of that many times. I don't have enough nodes for the HA part, so, I've tried to make do with 6 physical machines, each with at least two nics. | 17:33 |
kurt_ | kentb: I'm doing mine completely on VMs | 17:34 |
kurt_ | kentb: and you are running in to one of the challenges I am facing as well. I'm consolidating a lot of services to fewer nodes. | 17:35 |
kentb | kurt_: yeah, I'm wondering if putting too much on one machine might be hurting me | 17:35 |
kentb | (with juju-core) | 17:36 |
kurt_ | these guys test with a small amount of nodes, so they have managed to figure it out | 17:36 |
kurt_ | if there was one thing I wish were out there (hint hint jcastro) | 17:36 |
kurt_ | it would be a minimal install blueprint | 17:36 |
kurt_ | I hope to figure that out on my own with what knowledge I have | 17:37 |
kentb | me too! | 17:37 |
kurt_ | and eventually I am going to create a guide with all of this info - but I'm out in the wild right now | 17:37 |
kentb | join the club :) I feel like I'm really close. | 17:38 |
kurt_ | keep checking in to this channel. Since we are on the same track, we should share ideas | 17:38 |
kentb | will do! | 17:39 |
kentb | and I agree | 17:39 |
kurt_ | these guys do appreciate the bugs you find, so keep the info flowing | 17:39 |
kentb | definitely! | 17:39 |
jcastro | kurt_: you mean for openstack? | 17:40 |
kurt_ | yes sir | 17:40 |
jcastro | yeah | 17:40 |
kentb | yep | 17:40 |
jcastro | adam_g: I'm supposed to talk to you about an openstack bundle actually | 17:40 |
kentb | quantum-gateway is kicking my butt | 17:40 |
kurt_ | that is my numero uno goal right now | 17:40 |
kurt_ | to get a working openstack deployment on VMs with the minimum install | 17:41 |
kurt_ | but not the "virtual" method | 17:41 |
kurt_ | I would love to have a "scalable" blueprint working with maas, juju, juju-gui and openstack | 17:42 |
marcoceppi | kurt_: There are a few deployer configs out there for that | 17:42 |
kurt_ | marcoceppi: yup, but I've not been successful yet in making one work | 17:43 |
kurt_ | I've gotten close, but no cigar | 17:43 |
kentb | same here | 17:43 |
=== CyberJacob|Away is now known as CyberJacob | ||
jcastro | yeah, so arosales told me this morning that adam_g's been working on a working bundle | 17:43 |
marcoceppi | kurt_: kentb thanks for sticking with it. We're working on making this a very strong story in the near future | 17:43 |
kurt_ | marcoceppi: I know you guys are. I believe you feel this is an important and compelling user case and story | 17:44 |
kurt_ | that's why I've been working hard on this | 17:44 |
kentb | marcoceppi: my pleasure. I'm learning a *ton* I'm also working with a big OEM on a whitepaper on how to do this on their hardware. | 17:44 |
kentb | and I drew the openstack straw :) | 17:45 |
kurt_ | marcoceppi: can you share those deployer configs you spoke of to see if there is anything I don't know about? | 17:47 |
marcoceppi | adam_g: where's be most recent version of the openstack deployers? they still in the deployer repo? | 17:48 |
* marcoceppi knows you're working with jcastro on this as well | 17:48 | |
jcastro | I only found out today | 17:48 |
jcastro | but I am keen on getting my hands on whatever he has, heh | 17:48 |
* marcoceppi thinks we all are ;) | 17:48 | |
adam_g | marcoceppi, in lp:openstack-ubuntu-testing, juju-deployer -c etc/deployer/deployments.cfg -l | 17:48 |
kurt_ | are these similar to devstack type things? | 17:49 |
adam_g | those are all sort of specific to our lab, using custom charm branches and some lab-specific config | 17:49 |
adam_g | gimme a minute and ill put together a vanilla one for a simple deployment | 17:49 |
kurt_ | adam_g: nice one, thanks mate | 17:49 |
kentb | woohoo | 17:50 |
marcoceppi | kurt_: so juju-deployer, if you're not aware, is a means to standup complicated juju environments | 17:50 |
jcastro | adam_g: hey so, I am thinking make a simple one, post it on the list, and ask for feedback | 17:50 |
marcoceppi | it's just a yaml file that can be used with juju-deployer to deploy, configure, and relate services | 17:50 |
adam_g | jcastro, yeah, i need to put up a wiki that documents this. i have a WI for it and hope to get to it soon. | 17:50 |
kurt_ | marcoceppi: I saw that information for the first time at the bottom of jamespage's manifesto | 17:50 |
kurt_ | marcoceppi: is the juju-deployer discussed at length anywhere? | 17:52 |
jcastro | hazmat: maybe it's time to post on the list about deployer as well | 17:54 |
kurt_ | I would love to read about what it does and how it does it | 17:54 |
adam_g | jcastro, http://paste.ubuntu.com/6093510/ | 17:54 |
adam_g | jcastro, this is a vanilla openstack + ceph. ceph node is single node (with no redundnacy), swift is single storage node as well | 17:55 |
jcastro | is this that deployments.cfg or is this a new thing? | 17:55 |
kurt_ | adam_g: how many nodes is this? is it single-node? | 17:56 |
adam_g | kurt_, every service in its own machine | 17:56 |
jcastro | ok so you could add-unit to this? | 17:56 |
kurt_ | how about the quantum-networking - is there a local.yaml or something it is referencing? | 17:57 |
jcastro | 2013-09-11 13:57:55 Deployment name must be specified. available: openstack-services', 'precise-grizzly', 'raring-grizzly') | 17:58 |
jcastro | which one do I use? | 17:58 |
kurt_ | I think both kentb and myself have managed to get pretty far, but have not been able to get a complete working set up because of the networking | 17:58 |
kentb | yeah, that backfires almost every time...there's something screwy with keystone in the end product that I'm not sure what broke | 17:58 |
jcastro | adam_g: where do you want the wiki page to be? I can start documenting this | 17:59 |
kurt_ | kentb: I don't think keystone is the problem | 17:59 |
kurt_ | at least in my set ups it wasn't | 17:59 |
kurt_ | it was the ability to assign an IP and spin up vm's from horizon | 18:00 |
kentb | kurt_: really? What were you hitting? For me, I was always getting a 401 error if I tried to do anything with quantum. | 18:00 |
kurt_ | which was probably due to my networking being incorrect | 18:00 |
adam_g | jcastro, im not sure. somewhere near the current openstack HA wiki page? dont have URL handy | 18:00 |
jcastro | jujuclient.EnvError: <Env Error - Details: | 18:01 |
jcastro | { u'Error': u'invalid entity name or password', | 18:01 |
jcastro | u'ErrorCode': u'unauthorized access', | 18:01 |
adam_g | jcastro, https://help.ubuntu.com/community/UbuntuCloudInfrastructure | 18:01 |
jcastro | I get this kind of stuff when I try to run deployer on that pastebin'ed bundle | 18:01 |
jcastro | https://help.ubuntu.com/community/UbuntuCloudInfrastructure/JujuBundle for now | 18:02 |
jcastro | adam_g: ok, so now, next step is what to do wrt. openstack-service, precise-grizzly, or raring-grizzly | 18:04 |
jcastro | I have an environment up and running | 18:05 |
jcastro | juju-deployer -c openstack.cfg -elocal precise-grizzly | 18:06 |
jcastro | this seems right? Looks like it's working | 18:06 |
kurt_ | jcastro: are you able to access horizon and spin up VMs? | 18:06 |
jcastro | it's firing up right now | 18:06 |
jcastro | gimme like 5 minutes | 18:06 |
kurt_ | kk | 18:06 |
jcastro | juju-deployer -v -c openstack.cfg -elocal precise-grizzly | 18:07 |
adam_g | jcastro, good luck using local provider :) | 18:07 |
jcastro | adam_g: I just wanted to get the syntax for the command down, etc. | 18:08 |
adam_g | ah, right | 18:08 |
adam_g | jcastro, openstack-services is the base deployment, precise-grizzly just inherits and sets series and the config to install grizzly | 18:08 |
jcastro | adam_g: can you add some info here? https://help.ubuntu.com/community/UbuntuCloudInfrastructure/JujuBundle | 18:11 |
adam_g | jcastro, on a call atm. i will, cant promise its going to happen this week tho | 18:11 |
jcastro | k | 18:11 |
jcastro | do you think this will fire up on like hpcloud or something? | 18:11 |
jcastro | I'd like to see it work at least once! | 18:12 |
adam_g | not sure | 18:12 |
jcastro | but it works on MAAS? | 18:13 |
kurt_ | adam_g: if you could include some info around how the networking is handled (i.e. IP/CIDRs, interfaces, public ranges, etc) , I would be grateful | 18:13 |
kurt_ | jcastro: did VM spin up? | 18:17 |
jcastro | a bunch of containers spun up | 18:17 |
jcastro | fails on swift-storage-z1 | 18:17 |
kurt_ | anything interesting in debug-log that's an easy fix? | 18:18 |
jcastro | http://imgur.com/Bh114KN | 18:19 |
jcastro | this is the 2nd time I'm trying it, I'll check debug log | 18:19 |
jcastro | failing on the install hook | 18:19 |
kurt_ | it looks like most of them are stuck deploying | 18:21 |
jcastro | yeah, a bunch of them are turning green now | 18:21 |
kurt_ | nice | 18:21 |
kurt_ | can you pastebin your install hook error for swift? | 18:22 |
jcastro | ok so deployer showed an error | 18:22 |
jcastro | but the unit came up just fine | 18:23 |
jcastro | and I think I found a bug in nova-compute though | 18:23 |
jcastro | adam_g: on the nova-compute charm: http://pastebin.ubuntu.com/6093626/ | 18:23 |
jcastro | should I report that as a bug? | 18:24 |
kentb | ah! that might be what's killing me too...my nova-compute instance was DOA and libvirt was all messed up as one of the symptoms | 18:24 |
adam_g | jcastro, full log? i believe thats one of the many issues you'll hit doing this in containers | 18:24 |
jcastro | http://paste.ubuntu.com/6093637/ | 18:26 |
kurt_ | adam_g: wasn't log filling issue fixed in 1.13? | 18:35 |
kurt_ | (juju 1.13) | 18:35 |
kentb | kurt_: yep...hasn't come back for me since updating | 18:35 |
kentb | bug killed my bootstrap node within a few hours | 18:36 |
kurt_ | yes, mine too in 1.12 | 18:36 |
=== defunctzombie_zz is now known as defunctzombie | ||
kentb | ok. so if an instance is stuck in 'dying' state is there a good way to nuke it? I can't terminate the machine b/c we're indefinitely stuck there. Please tell me I don't have to destroy-environment and start over (using juju-core 1.13.3-1-1737). | 20:06 |
kentb | the agent-state is 'error' with a hook-failure during config-changed | 20:06 |
kentb | nm I ran juju resolved ceph and then that allowed me to kill it | 20:13 |
thumper | jcastro: ping | 20:25 |
jcastro | yo! | 20:26 |
thumper | jcastro: got a few minutes for a hangout? | 20:26 |
jcastro | yeah let me finish something | 20:27 |
jcastro | ~10 min? | 20:27 |
thumper | sure | 20:27 |
kurt_ | kentb: juju resolved <service> | 20:44 |
kurt_ | kentb: juju resolved <service> | 20:45 |
kurt_ | I just went over this with marcoceppi yesterday | 20:45 |
marcoceppi | kurt_: yes, we need to update the documentation for this | 20:46 |
kurt_ | Can anyone tell me if there is default username/password for console only access for nodes? I don't have ssh access | 20:46 |
marcoceppi | kurt_: everything is done via ssh | 20:47 |
kurt_ | I'm hosed them | 20:47 |
marcoceppi | kurt_: you can try ubuntu with the password ubuntu | 20:47 |
kurt_ | then | 20:47 |
kurt_ | that doesn't work | 20:47 |
marcoceppi | then I don't think so | 20:47 |
marcoceppi | We try to avoid default passwords and users, because that's a vulnerability | 20:48 |
kurt_ | macroceppi: if I am trying to add an interface after the fact to node? I keep messing up my routing when I do it | 20:48 |
kurt_ | sorry, that wasn't clear | 20:48 |
kurt_ | is there an easy way to add an interface to a node once it's been deployed? | 20:48 |
kurt_ | I appear to keep screwing up my routing | 20:49 |
kurt_ | this is on maas btw | 20:49 |
kurt_ | not that that matters | 20:49 |
kentb | kurt_: yep, that unclogged it. thanks! | 20:52 |
kurt_ | kentb: good stuff | 20:53 |
jcastro | hey thumper | 20:55 |
marcoceppi | kurt_: no, you can't add interfaces after the fact unless you use upgrade-charm | 20:56 |
kurt_ | upgrade-charm? | 20:56 |
marcoceppi | kurt_: yes. So if you're developing a charm locally, and you deploy using --repository and local:, and you later ADD a relation/interface to the metadata.yaml, you can run `juju upgrade-charm --repository ... <service>` to upgrade the charm and register the new relation/interface | 20:57 |
marcoceppi | kurt_: chaning interfaces/relations or removing them can be quite dangerous if you don't first remove all relations | 20:58 |
* marcoceppi is working on the upgrade-charm docs atm | 20:59 | |
kurt_ | so, do I have to entertain the idea of completely statically assigning my maas installation to make openstack work as in jamepage's doc? | 20:59 |
kurt_ | I am sure I could avoid using the vip parameter, but does that only apply to HA situations? Or is it a real virtual IP that can be assigned on top of juju's administrative IP? | 21:02 |
kurt_ | Looks like this is going to be a problem too: https://bugs.launchpad.net/juju/+bug/1188126 | 21:11 |
_mup_ | Bug #1188126: Juju unable to interact consistently with an openstack deployment where tenant has multiple networks configured <canonistack> <openstack> <serverstack> <juju:New> <juju-core:Triaged> <https://launchpad.net/bugs/1188126> | 21:11 |
=== CyberJacob is now known as CyberJacob|Away | ||
=== CyberJacob|Away is now known as CyberJacob | ||
=== kentb is now known as kentb-out | ||
=== freeflying_away is now known as freeflying | ||
=== CyberJacob is now known as CyberJacob|Away | ||
=== defunctzombie is now known as defunctzombie_zz | ||
=== defunctzombie_zz is now known as defunctzombie | ||
=== defunctzombie is now known as defunctzombie_zz | ||
freeflying | how many constraints from juju python has been implemented in juju-core? | 23:34 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!