=== mattgrif_ is now known as mattgriffin | ||
=== kadams54 is now known as kadams54-away | ||
=== scuttlemonkey is now known as scuttle|afk | ||
apuimedo_ | hi all | 09:17 |
---|---|---|
apuimedo_ | lazyPower: stub: About the problem with cs:precise/cassandra where deploying to an lxc is problematic due to the lack of hostname resolution | 09:18 |
apuimedo_ | the more I think about it, the more I think this is a problem of how juju deploys lxc | 09:18 |
apuimedo_ | http://paste.ubuntu.com/10865222/ | 09:21 |
apuimedo_ | following http://www.debian.org/doc/manuals/debian-reference/ch05.en.html#_the_hostname_resolution | 09:22 |
apuimedo_ | the lxc deployment misses the identity between 127.0.1.1 and the hostname | 09:22 |
=== scuttle|afk is now known as scuttlemonkey | ||
lazyPower | apuimedo_: i think this goes further than just juju however - sudo lxc-create -t ubuntu -n awesome -- only names the container awesome. The underlying container doesn't get the hostname specified on the CLI | 12:57 |
lazyPower | apuimedo_: so it appears to me that this is bugworthy against LXC with a follow up for juju - but i'm not sure thats intended behavior either. | 12:57 |
apuimedo_ | it does get a hostname though | 12:57 |
apuimedo_ | juju just does the lxc-create? | 12:58 |
lazyPower | hmm let me re-check... i am bleary eyed and sipping my first cup of coffee | 12:58 |
apuimedo_ | ;-) | 12:58 |
lazyPower | yeah, it basically just clones a container template and does a few setup steps | 12:58 |
apuimedo_ | better go with "earl grey, hot!" | 12:58 |
lazyPower | apuimedo_: ah i see what you're saying | 12:59 |
lazyPower | it sets /etc/hostname but doesn't update /etc/hosts with the loopback resolution | 12:59 |
apuimedo_ | exactly | 13:00 |
lazyPower | yeah - again - i point @ lxc for not doing that yet. i feel its bugworthy | 13:00 |
apuimedo_ | it's an inconsistency with what happens for normal images like you'd have in OSt/maas | 13:00 |
apuimedo_ | lazyPower: do you have a bugtracker link handy? | 13:01 |
lazyPower | https://bugs.launchpad.net/ubuntu/+source/lxc/+filebug | 13:01 |
apuimedo_ | thanks ;-) | 13:03 |
lazyPower | apuimedo_: did you land a branch MP against the legacy cassandra charm to update teh hostname? | 13:05 |
lazyPower | s/land/propose/ | 13:05 |
apuimedo_ | lazyPower: https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/1447160 | 13:08 |
mup | Bug #1447160: lxc-create template does not include /etc/hosts hostname resolution <lxc (Ubuntu):New> <https://launchpad.net/bugs/1447160> | 13:08 |
apuimedo_ | lazyPower: no, not yet. I was first checking the source of the problem | 13:08 |
lazyPower | Sounds reasonable. :) Thanks for the link, i subbed | 13:09 |
apuimedo_ | do you think it'd be acceptable to just do echo "127.0.1.1 $(cat /etc/hostname)" >> /etc/hosts as a first step in the charm install if grepping the /etc/hosts for the hostname fails? | 13:11 |
lazyPower | Sure | 13:12 |
lazyPower | i'd want to test it, but that sounds reasonable | 13:12 |
jamespage | gnuoy, https://code.launchpad.net/~james-page/charms/trusty/rabbitmq-server/vivid-fixes/+merge/257084 | 13:17 |
jamespage | following charm-helpers for that one | 13:17 |
gnuoy | jamespage, +1 | 13:21 |
jamespage | gnuoy, https://code.launchpad.net/~james-page/charms/trusty/mongodb/vivid-fixes/+merge/257086 | 13:22 |
gnuoy | jamespage, lots of conflicts with that one | 13:22 |
jamespage | gnuoy, opps - try again - https://code.launchpad.net/~james-page/charms/trusty/mongodb/vivid-fixes/+merge/257086 | 13:24 |
gnuoy | jamespage, +1 | 13:27 |
dosaboy | jamespage, gnuoy: 3-phase has yeilded better thus far | 13:29 |
gnuoy | excellent | 13:29 |
dosaboy | jamespage, gnuoy: we hit an issue with swift/ceph on one nonde but we think it is down to hw issue | 13:29 |
dosaboy | jamespage, gnuoy: sgdisk gets stuck in D-state | 13:29 |
jamespage | gnuoy, charm support for lxd - https://code.launchpad.net/~james-page/charms/trusty/nova-compute/lxd/+merge/257104 | 14:02 |
mwak | hi | 14:24 |
marcoceppi_ | o/ | 14:24 |
mwak | any idea why I get the following error when deploying a node app | 14:24 |
mwak | http://pastebin.com/bL4CuRV5 | 14:24 |
mwak | ? :) | 14:24 |
marcoceppi_ | mwak: Juju seems to think you don't have an available machines | 14:25 |
marcoceppi_ | any "clean" machines | 14:26 |
marcoceppi_ | even though you clearly do | 14:26 |
mwak | hum | 14:33 |
mwak | weird :/ | 14:33 |
marcoceppi_ | mwak: it could also be a default-consrtaints issue | 14:34 |
marcoceppi_ | mwak: can you run `juju get-constraints`? | 14:34 |
mwak | no constraints | 14:34 |
marcoceppi_ | mwak: ah | 14:34 |
marcoceppi_ | I see the issue | 14:34 |
marcoceppi_ | mwak: you're deploying a precise charm, you only have trusty machines allocated | 14:35 |
marcoceppi_ | juju is trying to add another machine to match this new constraint of series | 14:35 |
marcoceppi_ | and fails | 14:35 |
jcastro | ah, so simple it's obvious | 14:35 |
mwak | oh, no node-app for trusty | 14:36 |
mwak | :( | 14:36 |
marcoceppi_ | mwak: not atm :( | 14:36 |
marcoceppi_ | you can probably fork it, push to persoanl branch in charm-store, then deploy from there | 14:37 |
marcoceppi_ | it shoudl work int trusty, it just needs integration tests to move forward | 14:37 |
mwak | yup | 14:37 |
mwak | will do that | 14:39 |
stokachu | sinzui: The Juju AWS and MAAS providers now support starting LXC containers. | 14:49 |
stokachu | sinzui: does that mean lxc containers will pull from and managed MAAS dns/dhcp? | 14:49 |
stokachu | a* | 14:49 |
sinzui | stokachu, container networking is SNATing the container to the host machine. Though dimitern cab explain specifics | 14:51 |
stokachu | be interesting to know if those containers are reachable without having to go through juju ssh | 14:51 |
apuimedo_ | lazyPower: https://code.launchpad.net/~celebdor/charms/precise/cassandra/hostname_resolve/+merge/257120 | 14:56 |
dimitern | stokachu, it depends, so in 1.23.2 with the "address-allocation" feature flag on, we'll use maas api to allocate static ips for containers | 15:10 |
stokachu | dimitern: sweet | 15:10 |
dimitern | stokachu, if the feature flag is not on, juju-br0 will be used with DHCP for both kvm and lxc on maas | 15:10 |
stokachu | ah ok understood | 15:11 |
stokachu | that is pretty sweet | 15:11 |
dimitern | stokachu, we'll most likely change the address allocation with juju-br0 to use the maas devices api to tell maas as juju starts containers on a node (so maas can know which device belongs to which parent node and cleanup dhcp leases, etc.) | 15:12 |
dimitern | but that will happen when 1.8 is in trusty | 15:12 |
stokachu | dimitern: very cool | 15:13 |
dimitern | stokachu, I'm composing a reply to the mail sinzui sent with your questions | 15:19 |
sinzui | stokachu, the only change from the other 5 1.23.x releases was the feature flag. | 15:20 |
dimitern | sinzui, yeah, if you're not counting like 20-ish fixed issues :) | 15:28 |
=== urulama is now known as urulama|afk | ||
=== kadams54 is now known as kadams54-away | ||
=== dpb1_ is now known as dpb1 | ||
=== mattgrif_ is now known as mattgriffin | ||
=== kadams54 is now known as kadams54-away | ||
=== kadams54-away is now known as kadams54 | ||
=== kadams54 is now known as kadams54-away | ||
lamont | services framework relation, one direction I see the provided data show up on the other side. but not in the other direction.. anyone got a second to stare at it and call me silly? | 22:33 |
=== scuttlemonkey is now known as scuttle|afk | ||
=== arosales__ is now known as arosales |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!