=== CyberJacob is now known as CyberJacob|Away === Mike is now known as Guest21559 === cmagina-away is now known as cmagina === cmagina is now known as cmagina-away === vladk|offline is now known as vladk [06:33] marcoceppi: hey, just as a heads up, you may want to close this bug when/if the charm is on the store: https://bugs.launchpad.net/charms/+bug/1294232 [06:33] <_mup_> Bug #1294232: Charm Needed: Magento === vladk is now known as vladk|offline === psivaa_ is now known as psivaa-afk [09:12] marcoceppi: hi [09:17] if provider charm has only relation joined and requirer charm only relation changed will it be wrong? here provider relation joined sets variable and requirer charm gets those variables === CyberJacob|Away is now known as CyberJacob === CyberJacob is now known as CyberJacob|Away === vladk|offline is now known as vladk === CyberJacob|Away is now known as CyberJacob [11:32] hi, I tried manual providing yesterday and it was working like a charm [11:33] now if I try to connect to the agent machine agent (just doing juju status in fact) [11:33] ERROR state/api: websocket.Dial wss://my.domain.com:17070/: dial tcp XXX.XXX.XXX.XXX:17070: connection refused [11:34] s/manual providing/manual provisioning/ === CyberJacob is now known as CyberJacob|Away === vladk is now known as vladk|offline [12:55] Tug: you're using 1.16? [12:55] nop I switch to devel :s [12:56] devel should work with manual provider. what version Ubuntu is the machine? [12:58] the machine with the client or the agent ? [12:58] agent [12:58] ubuntu server 12.04 [12:58] :/ [12:59] yeah it was an experiment... [12:59] ssh in to the machine, run sudo initctl list | grep juju [12:59] so far you are doing all the right things [13:00] I have "juju-db stop/waiting" that's no good right ? [13:00] you should have two juju jobs [13:00] so the provisioning seems incomplete [13:01] jujud-unit-redis-master-0 start/running, process 10004 [13:01] jujud-unit-nirror-front-0 start/running, process 10006 [13:01] jujud-unit-nirror-proxy-0 start/running, process 10005 [13:01] juju-db stop/waiting [13:01] jujud-machine-0 start/running, process 10003 [13:01] but yes, that should be running [13:01] oh, start the juju-db job then [13:01] I may have reconfigured mongodb for my needs [13:01] is it using mongodb ? [13:01] sudo start juju-db [13:01] yes [13:01] :s [13:01] juju-db is mongodb [13:02] yeah I wiped out the db to connec to a replicaset [13:02] which explains why now. if you want to run a charm on node 0 which uses mongodb, try --to lxc:0 instead [13:03] ok but I would have to start over because the configuration is lost right ? [13:05] Tug: you might be able to save it but odds are it's best to start over [13:05] ok, thanks for helping me again marcoceppi :) [13:06] Tug: np! [13:07] next time I try the real thing in AWS ;) [13:23] ping marcoceppi [13:25] I wonder how I will update my application once deployed though. Is there an API or something to execute commands on all the units of a service ? [13:25] I like the way deployment is done with capistrano [13:47] I'm not a fan of polling a deployment repository === CyberJacob|Away is now known as CyberJacob === CyberJacob is now known as CyberJacob|Away === CyberJacob|Away is now known as CyberJacob [21:16] Tug: typically, you can make a version configuration option to decouple charm from application version [21:17] Tug: some charms even do something like repo, branch where repo can be a git, bzr, or hg repository and branch can be a git tag, git branch, bzr tag, etc [21:19] ok, but if I change the config option of the service all units are updated ? [21:19] what if I want to update half of it to keep the system online for instance ? [21:59] Tug: I may be wrong, but i don't think juju supports that ootb. You woudl want ot deploy 2 clusters of the application, and update them independently behind a load balancer [21:59] eg: wordpress1, wordpress2, behind haproxy [21:59] set wordpress1 to perform the update, haproxy will route to wordpress2 while wordpress1 is "down" for the update. === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [22:48] Tug: lazyPower is correct. You're managing service groups not units. If you want to matinain diffferent service groups and do cascading upgrades, deploy your service twice with different alias (juju deploy wordpress wp1; as an example) [22:48] ok makes sens :) [22:48] *sense [22:49] have to go, thx for tthe tips