=== natefinch-afk is now known as natefinch | ||
marcoceppi | stokachu: I've seen this, well something similar, before | 02:40 |
---|---|---|
redelmann | hello, im trying to remove a relation between service and subordinate in juju 1.23. It's posible? | 02:41 |
redelmann | juju destroy-relation says: ERROR relation "logstash-agent:juju-info ubuntu:juju-info" not found | 02:43 |
redelmann | but juju status show a relation | 02:43 |
mahmoh | marcoceppi: any idea for a fix? | 02:47 |
marcoceppi | redelmann: can you show your juju status? | 02:48 |
marcoceppi | mahmoh: I'm not sure, but it is likely a proxy issue fwiu | 02:48 |
mahmoh | marcoceppi: yeah, I agree just unsure what; I think it's in Juju since the deployed node can now download the lxc templates manually as sudo, sudo lxc-create ... and it works fine but fails with juju ... --to lxc:0 | 02:49 |
mahmoh | marcoceppi: I'll keep on it | 02:49 |
marcoceppi | mahmoh: have you tried juju retry-provisioning now? | 02:50 |
redelmann | marcoceppi, http://paste.ubuntu.com/11533810/ | 02:50 |
redelmann | redelmann, juju status is extra large, so i remove irrelevant parts | 02:51 |
mahmoh | marcoceppi: no, nothing changed: deployed env w/ lxc, lxc fails in juju, juju ssh # ... sudo lxc-create ... works | 02:51 |
redelmann | marcoceppi, juju status is extra large, so i remove irrelevant parts | 02:51 |
marcoceppi | well, logstash doesn't apper to be on ubuntu, which service are you trying to remove it from? juju-gui? fail2ban? | 02:51 |
=== X-Rob_ is now known as X-Rob | ||
rick_h_ | redelmann: all services have a relation with juju-info I believe. Any relation that starts with juju-* isn't exposed to users to modify/mess with | 02:55 |
marcoceppi | oh, duh, you also can't remove subordiantes at this time. | 02:56 |
rick_h_ | redelmann: https://jujucharms.com/docs/devel/authors-implicit-relations | 02:56 |
redelmann | marcoceppi, i deploy logstash to machine0.... big mistake | 02:57 |
* rick_h_ runs away now | 02:57 | |
redelmann | rick_h_: yes, i read that, i just dont understand why | 02:58 |
redelmann | ok, thank marcoceppi and rick_h_ . Bed time. | 02:59 |
redelmann | good bye | 03:00 |
=== freeflying__ is now known as freeflying | ||
merkurus | what does juju mean - literary | 04:43 |
blr | merkurus: a juju is a an object of spiritual/magical significance, from west african religions. | 04:47 |
blr | one too many articles there... | 04:48 |
merkurus | blr : aah | 04:56 |
=== blahdeblah_ is now known as blahdeblah | ||
jam | wallyworld: I'm available now, we'll need to keep it pretty quick since I'm at EOD and my son needs some ho | 13:03 |
jam | homework help | 13:03 |
jam | wallyworld: https://plus.google.com/hangouts/_/canonical.com/tanzanite ? | 13:03 |
wallyworld | ok | 13:03 |
pmatulis | if i want to add-unit will all the parameters of the existing service be used [ such as --constraints (tags) and --config (file) ]? if so, any way to see that metadata of existing service? | 14:56 |
Odd_Bloke | I have a place where I'm catching an exception in a charm, I'd like to log about it (ideally including traceback); is there any good way of doing that? | 15:58 |
Odd_Bloke | I notice that hookenv.log is just a function (i.e. I can't do hookenv.log.exception('...') as with normal Python logging). | 15:59 |
marcoceppi | Odd_Bloke: you could use the logging library, stdout is captured as DEBUG level output in the juju log for that service | 16:10 |
Odd_Bloke | Ah, shiny. | 16:13 |
nodtkn | juju add-machine ssh:root@junor1-1 returns "ERROR error checking if provisioned: subprocess encountered error code 1" | 17:01 |
nodtkn | with --debug I set that it creates the ubuntu user but then fails when checking to see if machine is already provisioned | 17:02 |
marcoceppi | nodtkn: was this machine at one point provisioned? | 17:02 |
nodtkn | marcoceppi: maybe, but I don't think so | 17:03 |
marcoceppi | nodtkn: if you log on to that machine, what does `initctl list | grep juju` show? | 17:03 |
nodtkn | marcoeppi: It does not show anything.... I asked around and found out that it has been provisioned before | 17:04 |
nodtkn | do I need to reinstall or is there a easy way to convince it that it is no longer provisioned | 17:05 |
=== kadams54_ is now known as kadams54-away | ||
=== kadams54-away is now known as kadams54_ | ||
marcoceppi | nodtkn: you may just need to do some clean up | 17:19 |
marcoceppi | nodtkn: is there a /var/lib/juju directory? | 17:19 |
Makyo | Trying to deploy a local charm (having successfully deployed it earlier today), and getting "ERROR invalid character 'w' looking for beginning of value" with no further info. Is there a way to dig into that a bit more? 1.24-beta5-utopic-amd64 | 18:04 |
marcoceppi | Makyo: check metadata.yaml or actions.yaml or config.yaml for typos | 18:07 |
nodtkn | marcoeppi: /var/lib/juju does not exist | 18:07 |
Makyo | marcoceppi, Everything looks good. Like I said, I deployed this morning fine, and now I'm getting that error. bzr diff/stat show no changes. | 18:09 |
mbruzek | marcoceppi: Is there a way to wait until a charm has started or juju is idle after a deploy? | 18:41 |
marcoceppi | mbruzek: inside a charm or from cmdline? | 18:41 |
mbruzek | marcoceppi from command line | 18:41 |
mbruzek | juju deploy ubuntu | 18:41 |
marcoceppi | mbruzek: if you use 1.24, yes | 18:41 |
mbruzek | wait_until_started | 18:41 |
mbruzek | lay it on me man | 18:41 |
marcoceppi | does it have to be bash? python would be faster | 18:42 |
mbruzek | python is fine | 18:42 |
mbruzek | amulet.Deployment.wait()? | 18:44 |
marcoceppi | mbruzek: no, that's the old way | 18:53 |
marcoceppi | I've got a better way if you're using 1.24 | 18:53 |
kwmonroe | juju deploy ubuntu && sleep 36000 && do_work | 18:54 |
kwmonroe | should be g2g mbruzek | 18:55 |
kwmonroe | hey friends, can i kick off a quickstart deployment, juju switch to another substrate in a new terminal, and kick off another? | 18:57 |
marcoceppi | kwmonroe mbruzek https://gist.github.com/marcoceppi/002ce8122f49a15c0153 | 19:03 |
mbruzek | marcoceppi: my computer froze up just before your answer, I got the link. | 19:19 |
mbruzek | marcoceppi: thanks | 19:20 |
kwmonroe | to answer my own question, yes, yes you can juju switch from another terminal and deploy to multiple substrates. good on juju! | 19:28 |
denverjoel | I installed juju-local on ubuntu 14.04.2 LTS. followed the directions on https://jujucharms.com/docs/stable/getting-started. when i get to the part about testing the wordpress / mysql setup, nothing pops up on the browser, or curling on the host machine, or while SSH'd into the juju machine. | 19:36 |
denverjoel | what else can i check to see if something is failing? | 19:37 |
kwmonroe | denverjoel: try running 'juju debug-log'. that should show you what the units in your setup are doing.. perhaps still installing packages? | 19:52 |
denverjoel | kwmonroe: this is my log http://pastebin.com/tR1ws0P8 | 19:56 |
denverjoel | it must be the wordpress charm because loading deploying an nginx charm works perfectly fine | 20:00 |
denverjoel | now i can access nginx via curl but still not externally. i have ufw enabled and tried disabled as well. there are iptables rules in place but all are set to accept. | 20:06 |
thumper | lazyPower: hey, did you see my email? | 20:08 |
kwmonroe | denverjoel: try 'juju status' and take a look at the wordpress service.. does it say "started" in that output? and/or have opened ports in that output? | 20:14 |
kwmonroe | also denverjoel, what version of juju are you using (juju version)? | 20:14 |
Bialogs | Is there a way to have multiple configurations attached to the same Juju charm? | 20:22 |
kwmonroe | denverjoel: double check that you did 'juju expose wordpress'.. juju status should show exposed: true in the output.. | 20:23 |
kwmonroe | Bialogs: you can deploy the same charm with different configs, if that's what you're after.. juju deploy --config=<config.yaml> <charm> | 20:25 |
denverjoel | kwmonroe: it is exposed but no open ports | 20:28 |
Bialogs | kwmonroe: No way to do that in the GUI? | 20:30 |
kwmonroe | ok denverjoel, and navigating to (or curling) the public_address for the wordpress service doesn't get you anything? (perhaps open ports arent listed in your version of juju) | 20:32 |
denverjoel | kwmonroe: curling inside the machine, outside of it: to localhost and the private IP doesn't yield anything. | 20:33 |
lazyPower | Bialogs: are you saying have 2 of the asme service stood up with different configurations? | 20:36 |
lazyPower | Bialogs: if thats the case, you will need to give them sep. service names, eg: juju deploy --config foo.yaml cs:trusty/mediawiki mediawiki-north | 20:36 |
lazyPower | and juju deploy --config bar.yaml cs:trusty/mediawiki mediawiki-south | 20:37 |
Bialogs | lazyPower: that's exactly what I'm looking for | 20:37 |
lazyPower | or once you've deployed the 2 services, edit the configs in the GUI - the key indicator is same service, different names for the service. | 20:37 |
lazyPower | service representations in juju essentially group the configuration for each unit that comprises the service | 20:37 |
lazyPower | so if you scale mediawiki-north, it inherets the config of the service for mw-north, same rule applies for mw-south | 20:38 |
denverjoel | how do i go about opening up ports? | 20:39 |
kwmonroe | denverjoel: juju expose wordpress should have done that for you. are you doing your deployment locally (lxc)? or with a cloud provider? | 20:41 |
denverjoel | local | 20:41 |
kwmonroe | hmm. in that case, open ports shouldn't even matter as you should be able to access container ports from your host machine willy nilly. | 20:43 |
denverjoel | im removing wordpress and mysql and re-deploying. | 20:43 |
kwmonroe | ack denverjoel, i'm gonna run through those steps locally too and see if i can repro. | 20:44 |
denverjoel | kwmonroe: immediately after the wordpress service started i curled and got 502 Bad Gateway. I retried curl and got nothing at all. | 20:46 |
denverjoel | strange, that was all without exposing the service. after exposing, i get no response from curl | 20:50 |
kwmonroe | ah, i think i know what's up denverjoel.. curl <private-ip> doesn't return anything because the index of http://<private-ip> just php includes some other stuffs. so i bet it's working, there's just nothing for curl to see there.. can you try "w3m <private-ip>" or lynx or elinks or even a real browser if you can? | 21:00 |
kwmonroe | for me, chrome'ing to 10.0.3.x shows the wordpress install wizard | 21:01 |
kwmonroe | but curl'ing that gives me nada. | 21:01 |
kwmonroe | that's without juju expose too, fwiw. | 21:01 |
=== kadams54_ is now known as kadams54-away | ||
Syed_A | hey guys, how do i remove a unit which is stuck in error state. juju remove-unit seems not to have any effect. | 21:22 |
thumper | Syed_A: you can move it out of error by going 'juju resolved unit-name/0' or whatever the unit name is | 21:23 |
thumper | then it should end up getting removed | 21:23 |
Syed_A | thumper: Thanks! It worked :) | 21:25 |
Syed_A | can i deploy heat in HA mode with juju ? | 21:51 |
* thumper shrugs | 21:56 | |
bleepbloop | jamespage: you around? | 22:01 |
bleepbloop | I am having an issue with config-changed failing for the nova-compute charm when trying to change virt-type, would anyone have any experience with this? | 22:08 |
Bialogs | What's the difference between gray and green connection lines in Juju | 22:44 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!