| === hml_ is now known as hml | ||
| kjackal_ | Good morning Juju world | 06:54 |
|---|---|---|
| junaidali | Guys I'm trying to bind shared-db on internal space, but it gives error "ERROR cannot add application "mysql": unknown space "public-api" not valid" | 07:11 |
| junaidali | here are my spaces define http://paste.ubuntu.com/23352337/ | 07:12 |
| junaidali | I'm running the following command to deploy juju deploy cs:xenial/percona-cluster mysql --config=openstackha.cfg --to lxd:13 --bind "shared-db=public-api | 07:14 |
| junaidali | sorry this one: | 07:14 |
| junaidali | juju deploy cs:xenial/percona-cluster mysql --config=openstackha.cfg --to lxd:13 --bind "shared-db=internal-api" | 07:14 |
| junaidali | is there any way to delete a space that we defined via juju? | 08:15 |
| junaidali | or update an existing space | 08:15 |
| aluria | hi -- on juju2 status output, what does an asterisk at the end of a unit name means? ie. "unitname/0* active idle ..." | 12:06 |
| magicaltrout | its the elected leader | 12:07 |
| aluria | ah, cheers | 12:08 |
| rick_h_ | bdx: :/ yea need to file a bug and get the UI team to chase it down. | 12:34 |
| rick_h_ | bdx: they might want to get an idea of how large your cookie files are and see what's up | 12:34 |
| === scuttle|afk is now known as scuttlemonkey | ||
| bryden | Hello, I just started using juju and previously I was able to load juju-gui on a local controller and access it through my browser no problem. For some reason all of a sudden it's now showing status "unknown" and workload unknown.. i can still access the gui through my browser but it hangs before I can login, saying "connecting to the juju model" with a spiraling circle. If anyone has any advice what I should investigate in order | 17:09 |
| bryden | to fix this please let me know! | 17:09 |
| === bryden is now known as bryden_ | ||
| bryden_ | a bit more info.. in the command line 'juju models' hangs as does 'juju controllers --refresh' | 17:51 |
| bryden_ | juju controllers shows me a controller but juju destrory-controller <controller-name> hangs | 17:52 |
| tvansteenburgh | bryden_: what version of juju? | 17:53 |
| bryden_ | 2.0-rc3-xenial-amd64 | 17:54 |
| tvansteenburgh | bryden_: was the controller created with that version? | 17:54 |
| bryden_ | @tvansteenburgh : 'juju controllers' returns controller with version listed as 2.0-rc3 | 17:56 |
| tvansteenburgh | bryden_: ok, not sure then. if it were me i'd upgrade to 2.0 stable and then bootstrap a new controller and use that | 17:59 |
| petevg | cory_fu, bcsaller: I rebased my feature/glitch-more-actions branch, deleted the remote, and repushed. New PR here: https://github.com/juju-solutions/matrix/pull/5 | 18:04 |
| petevg | If you've got the branch checked out locally, please delete it and refetch it. | 18:04 |
| petevg | (Apologies for the bit of git evil.) | 18:05 |
| hml | i have a charm whose config-changed hook is failing on read-only file system - trying to write to a file but that device is out of space - it feels like the issue if you don’t set up lxd correctly with zfs - however i’m using dir with lxd not zfs - any ideas? | 18:11 |
| bryden_ | tvansteenburgh ok, thanks I'm trying that out now | 18:11 |
| devops | hi everyone, can we change dns-name being displayed in juju status | 18:12 |
| devops | dns-ip* | 18:13 |
| devops | public-address* | 18:13 |
| bryden_ | tvansteenburgh: juju upgrade-juju --debug complains that no model is in focus | 18:15 |
| rick_h_ | devops: no, it's not mallable from the outside after the fact. | 18:15 |
| rick_h_ | bryden_: juju switch controller? | 18:15 |
| bryden_ | rick_h_ hangs | 18:15 |
| rick_h_ | bryden_: try with a --debug and see if it's got anything interesting in there? | 18:16 |
| bryden_ | https://paste.gnome.org/pip5jsct0 | 18:16 |
| bryden_ | last line keeps repeating | 18:17 |
| rick_h_ | bryden_: are you able to reach that network? | 18:18 |
| rick_h_ | bryden_: not sure about your setup there, but seems like a network issue not able to reach the controller | 18:18 |
| rick_h_ | bryden_: or that the controller went down for some reason, can you ssh to the controller machine? | 18:18 |
| bryden_ | rick_h_: ssh to the controller gives me connection refused | 18:19 |
| rick_h_ | bryden_: is this maas or something? | 18:20 |
| rick_h_ | bryden_: I would expect the machine to be down in that case | 18:20 |
| rick_h_ | bryden_: so curious if someone pulled a plug on you or something | 18:20 |
| bryden_ | rick_h_: this is all on my localhost, | 18:20 |
| rick_h_ | bryden_: so the lxd provider? | 18:20 |
| rick_h_ | bryden_: can you run sudo lxc list | 18:21 |
| rick_h_ | bryden_: and see the status of the containers there? | 18:21 |
| bryden_ | https://paste.gnome.org/pwwmcia6s | 18:22 |
| bryden_ | rick_h_: also this is the logfile for lxd not sure if anything from this is interesting... | 18:22 |
| bryden_ | https://paste.gnome.org/ptldqvfzi | 18:23 |
| rick_h_ | bryden_: so that seems to say you've got no lxd containers running which means whatever you had was removed | 18:24 |
| rick_h_ | bryden_: so the only thing from here would be to juju unregister the controller and remove it from your list of controllers and rebootstrap | 18:25 |
| bryden_ | rick_h_: working on a new bootstrap now thanks | 18:31 |
| rick_h_ | bryden_: let me know if you can see what happened there. The only two ways those containers should go away is if you destroy the controller or manually remove them with lxc tools | 18:31 |
| hml | rick_h_: hi. a while back i had an issue where I setup zfs for lxd by accident and ran out of space on charm units - i have a new issue that feels like that one - with openstack-novalxd bundle - lxd/0 is finding a readonly filesystem on /sys - however i double check not using zpool - any ideas? | 18:36 |
| rick_h_ | hml: so I know we landed a fix that tried to cap things at 90%, however if you were attacking too fast then it was still possible to trigger. | 18:37 |
| rick_h_ | hml: but should be much harder to do | 18:37 |
| rick_h_ | hml: as far as sys being read-only I'm not sure. It seems odd that only /sys would be read-only? is the whole filesystem read-only or just sys is on a different partition/etc? | 18:37 |
| hml | rick_h_: i’m deploying the openstack-novalxd bundle | 18:37 |
| hml | rick_h_: here’s the error message from the unit log: config-changed IOError: [Errno 30] Read-only file system: '/sys/module/ext4/parameters/userns_mounts' | 18:38 |
| rick_h_ | hml: oh hmm, so no zfs here? ext4? | 18:38 |
| hml | rick_h_: I hope so - i ran zpool list and the containers didn’t show up. | 18:39 |
| hml | rick_h_: only lxd/0 seems to be having a problem | 18:40 |
| bryden_ | rick_h_: I'm back to the beginning of my issue.. i did juju deploy juju-gui; juju expose juju-gui and when i go to the public address in my browser it just hangs with a turning circle and a message: "Connecting to the juju model". | 18:43 |
| rick_h_ | bryden_: k, can you juju status --debug and connect? | 18:43 |
| bryden_ | https://paste.gnome.org/ppcjs5unn | 18:45 |
| === plumgrid is now known as Baqar | ||
| rick_h_ | bryden_: is there a reason you're deploying the gui vs using the built in gui? | 19:04 |
| bryden_ | rick_h_: ignorance of the builtin gui | 19:05 |
| rick_h_ | bryden_: ah ok, guessing the charm gui might need some love or something | 19:06 |
| rick_h_ | bryden_: but running juju gui should get you one | 19:06 |
| rick_h_ | bryden_: it's built in these days so you don't need to deploy it | 19:06 |
| bryden_ | rick_h_:that's brilliant | 19:06 |
| bryden_ | rick_h_: and it works | 19:06 |
| bryden_ | rick_h_: thank you for spending time with me working on this | 19:07 |
| rick_h_ | bryden_: awesome, glad that's useful and let us know how it goes from there | 19:16 |
| === bryden is now known as bryden_ | ||
| === bryden is now known as bryden_ | ||
| === hml_ is now known as hml | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!