=== scuttle|afk is now known as scuttlemonkey [02:02] pmatulis: the difference is how juju models it, juju deploy ubuntu will create a machine with the ubuntu charm on it and the service ubuntu in your topology, juju add-machine will simply provoison a machine with no services deployed to it making it available for services to be deployed to it. [02:13] marcoceppi: yeah i'm starting to see it better now. thanks for the description [03:50] if anyone could help with my vivid-kilo, maas set up that would be awesome. I am running into a ceph issue http://askubuntu.com/questions/668320/ceph-cannot-reformat-because-disk-is-mounted [12:55] jamespage, looks like we need something like SWIFT_CODENAMES in charm-helpers for all the projects in liberty. I can start on that if you haven't. [13:05] coreycb, that should already be covered I think [13:05] but maybe not [13:08] jamespage, I think get_os_version_codename() and openstack_upgrade_available() need some updates [13:54] beisner, are you ware of any trusty/kilo issues? seeing a instance boot error on a mp I'm reviewing? [13:57] hi jamespage - t-k next mojo spec from yesterday fired up and connected to a new instance ok. http://paste.ubuntu.com/12253397/ or http://10.245.162.77:8080/view/Dashboards/view/Mojo/job/mojo_runner/745/console [13:58] beisner, this is the MP - https://code.launchpad.net/~bbaqar/charms/trusty/nova-cloud-controller/next/+merge/269897 [14:01] jamespage, digging into the logs collected on that run, nova-compute log shows MessagingTimeout and nova conductor woes. http://paste.ubuntu.com/12253431/ [14:05] jamespage, fyi Aug23: n-c-c/next most recent time-triggered, successfully completed run. (the Aug30 runs were sabotaged by undercloud storms.) [14:05] jamespage, Aug23 for reference: http://10.245.162.77:8080/view/Dashboards/view/Amulet/job/charm_amulet_test/5981/ [14:05] jamespage, i've re-triggered a new n-c-c/next amulet run now. [14:11] beisner: did any of those MP's come in lastnight? [14:11] beisner: i didn't get notice... or i did and they were filtered so i had no idea i had stuff to do [14:12] gnuoy, hey - could you give https://code.launchpad.net/~james-page/charm-helpers/plumgrid/+merge/269920 a quick +1 (its for the plumgrid SDN stuff - managed to miss they have not proposed to ch's) [14:13] jamespage, sur [14:13] * sure [14:13] ta [14:25] hi lazyPower - apologies if i kept ya hanging - that train is delayed, but could be arriving soon. [14:33] any openstack charmers on? [14:34] beisner: no stress. Just wanted to circle back and make sure I didn't leave you hangin lastnight :) [14:34] firl: do you have a question? [14:34] marcoceppi: Yeah, someone helped me get a config ready for vivid-kilo, and I am running into an issue with ceph, and wanted to know if it was me doing something wrong or something else [14:35] http://askubuntu.com/questions/668320/ceph-cannot-reformat-because-disk-is-mounted [14:35] beisner jamespage gnuoy coreycb ^ ? [14:36] firl: thanks for opening an Ask Ubuntu question on it [14:36] np, yeah it was early morning, and didn’t know if it should be a bug or a question [14:36] firl, hmm [14:37] firl: question is always a good start, it allows a medium to track history and gives people who arent' always online the ability to readback from there it might be a bug or just a fix to your environment [14:37] marcoceppi: good to know, thanks [14:38] firl, is this a re-deployment? [14:38] nope [14:38] i.e. have you managed to deploy once, and you're now re-deploying on the same hardware? [14:39] I am redeploying on same hard ware [14:39] but I am using mass to juju destroy-environment [14:39] I thought that the osd-reformat would take care of the designation on the newly deployed os ( on the same old hardware ) [14:40] firl: its possible that something systemd-ish is causing a race on the redeployed system [14:40] i.e. a udev rule fires and mounts the volume at about the same time this process is running [14:40] of where it is mounting the drive before ceph allocates it? [14:40] maybe - tricky to etll [14:40] gotcha [14:41] firl, ceph used udev rules to mount and start osd on reboot - its possible this is the cause [14:41] although I've not seen this one before [14:41] yeah [14:41] I have reproduced it 3 times [14:41] urgh [14:41] :) [14:41] I try to test before asking haha [14:41] firl, did you mean to deploy on vivid btw? [14:41] or was kilo the main requirement? [14:41] Kilo is the main [14:41] vivid-kilo was for LXD possibility [14:42] firl, ah! [14:42] I see [14:42] I can do trusty-kilo and see if it reproduces it [14:42] makes sense [14:42] :) [14:42] Makes me feel a little less crazy then haha [14:42] firl, I would probably bet its fine on trusty/kilo - that gets alot of redeployment exercise on hardware in test labs and at customer deployments [14:43] vivid - less so - so this type of thing can bite. [14:43] gotcha [14:43] so trusty/kilo is more along the lines of trusted / stable ? [14:43] firl: absolutely - but we don't provide LXD back to trusty ... [14:44] so I can see why vivid was making sense for you [14:44] yeah, yeah the LXD was just for fun testing [14:44] not an immediate need so to speak [14:44] firl, that said its probably worth highlighting that block device support is not yet implemented in LXD/nc-lxd [14:44] so cinder/ceph + LXD == ERRNOTSUPPORTED right not [14:44] ah that is good to know [14:44] yeah I use cinder/ceph so that makes sense [14:45] ok off to trusty-kilo [14:45] that is being worked on - but there are a number fo security concerns with work through in the kernel todo with fs integrity [14:46] very true [14:47] Do you know of any other customer type example config.yaml for the OS world to compare my set up against by any chance? [14:49] firl, it looks like you pulled yours from openstack-charm-testing right? [14:49] ya [14:50] firl, it you wanted something a little more concise to work from - https://jujucharms.com/openstack-base/ [14:50] that uses charm-store rather than branches and could easily be adapted [14:50] thanks man, I appreciate it [14:52] firl, that also uses the new bundle format which expresses everything about the environment including the physical machines [14:52] gnuoy, sorry to nag on that review - pretty please :-) [14:54] jamespage, sorry, it's just possible I got distracted by something shiny [14:54] ha [14:55] shiny? where?! [14:57] ddellav, hey - I see you glance upgrade action stuff got landed - nice work [14:57] gnuoy, ta [15:00] jamespage, the format that I am using? or the openstack-base does? [15:01] firl, the openstack-base bundle uses the new format [15:01] gotcha, yeah I will convert over to the new format and try to specify it [15:01] thanks again [15:04] firl, its worth noting that by using the charm-store for the charms, you can fix charm revisions for repeatable deployments [15:04] firl, worth doing to avoid an unexpected re-deployment change [15:04] say a new feature or suchlike [15:07] lazyPower, plumgrid changes for openstack charms all landed now [15:07] lazyPower, well into next at least [15:07] https://media.giphy.com/media/OJdyjK11k2fLi/giphy.gif [15:08] awesome, I'll clean up the technical debt i left behind before EOW and cc you when i get the bundle corrected. [15:08] thanks for the update jamespage [15:10] lazyPower, thanks for the final reviews whilst i was away - appreciated! [15:11] jamespage: np, happy to see them land their charms. :) [15:12] indeed [16:10] rbasak: I've got a debian packaing question, you got a min? [16:25] jamespage, thanks :) I've also landed the cinder changes as well. Working on ceilometer now. [16:26] jamespage, https://code.launchpad.net/~ddellav/charms/trusty/cinder/upgrade-action/+merge/269247 [16:26] feel free to take a look and comment. [16:26] ddellav, good-oh! [16:33] marcoceppi: sure. Also you just reminded me that I should have something on my todo for you I think. Do you remember what it is? [16:33] rbasak: I think I just found out what I did wrong, let me get back to you in a few mins [16:34] rbasak: so I cleaned up the charm-tools packaging. We left the discussion at "I need to do a 1.6.0 release so we can propose for archive) [16:34] " [16:34] marcoceppi: ah OK. So no action for me right now? [16:34] however, 1.6.0 added a bunch of deps not packaged that I just wrapped up packaging [16:34] rbasak: so I think after i finish testing/uploading to the PPA we should take another spin through the package to make sure everything is in place [16:35] I'll ping you tomorrow to chat more about [16:35] marcoceppi: OK. Let me know when you're ready. [16:35] ack, thanks [16:52] rbasak: okay, i actually do still have a packaging question [16:52] marcoceppi: shoot :) [16:52] * rbasak EODs soon [16:53] rbasak: so there was a source package, jujubundlelib and it installed jujubundlelib package. I wanted to create a python-jujubundlelib and python3-jujubundlelib package while making sure to maintain backwards compatibility with jujubundlelib package because others packages still depend on it [16:53] rbasak: so I created the following control file [16:54] https://github.com/juju/juju-bundlelib-packaging/blob/master/control [16:54] jujubundlelib depends on the python-jujubundlelib which replaces it, python-jujubundlelib breaks jujubundlelib 0.1.9-1 (this is now 0.1.9-2) and replaces jujubundlelib [16:55] however, if I isntall 0.1.9-1 and perform an upgrade I get this error [16:55] rbasak: http://paste.ubuntu.com/12254887/ [16:55] I have to run sudo apt-get install -f after the upgrade to have the package install properly [16:56] so I figured it was because I was dong << I patched to do <= but the same error occurs [16:56] 0.1.9-1 > 0.1.9-1~ [16:57] right so the control file is now [16:57] jujubundlelib (<= 0.1.9-1) [16:57] oh, so should i have done jujubundlelib (<< 0.1.9-2~) > [16:57] ?* [16:57] Yeah, jujubundlelib (<< 0.1.9-2~) is right [16:58] damnit, okay let me patch taht and try again [16:58] the ~ was throwing me off, I tried googling it but that's pretty ahrd [16:58] It's fixed in 0.1.9-2, so anything older than that is what it replaces. Suffixing ~ allows for backports and PPAs [16:59] cool [17:00] rbasak: thanks, I'll give this a go [17:00] Since if someone backports or PPAs 0.1.9-2 then it'll be in 0.1.9-2~, and that should also be treated as one where the transition has occurred. [17:00] No problem. [17:00] * marcoceppi adds a little more packaging knowledge to the tree === natefinch is now known as natefinch-afk [18:59] How do I make a failed hook as resolved? I am looking for an equivalent to `juju resolved` for a failed hook instead of a failed charm. [19:38] skylerberg: juju resolved unit/# resolves just that hook [19:38] if a follow up hook fails, the charm re-enters a failed state [19:40] lazyPower: Thanks. I got it working a bit ago, but I had tried several things, so I still wasn't sure what actually fixed it. [19:47] wolsen: I resubmitted my patch for nova-compute with the changes we discussed: https://code.launchpad.net/~sberg-l/charms/trusty/nova-compute/tintri-interface/+merge/269972. === natefinch-afk is now known as natefinch [19:49] skylerberg, ack - thanks - I'll take a look at it a bit later === danwest is now known as danwest-afk [20:43] wolsen, ping-o [20:51] beisner, pong-o [20:52] yo! [20:54] hey [20:54] wolsen, just touching base re: email (1c-h + 4 charm) review [20:56] wolsen, ah i think we left off with https://code.launchpad.net/~1chb1n/charm-helpers/amulet-svc-restart-race/+merge/269098 <-- rev pushed following review [20:57] beisner, yep I hadn't gotten back to your updated push on it [20:57] * wolsen looks now [20:59] wolsen, np. we had to let the tests run for the charms that i sync'd it into to confirm. [20:59] ie. those in your email consume these c-h changes [20:59] beisner, yep [21:00] wolsen, thx man [21:04] beisner, approved - added yet another comment, but its more of a follow up todo action [21:09] wolsen, replied fyi. another great question. [21:11] beisner, oh hah - my comment was misstated - heh, its ambiguous if len(pid_list) > 1 not 0 [21:11] beisner, but let me look at that method [21:11] wolsen, yes indeed, we are ignoring cases where there are multiple pids (as did the ancestor). which is a good to-do to further examine. [21:12] beisner, right - that was the crux of my comment - my mistake on the typo there - we both agree though ;) [21:12] beisner k, I don't have rights to merge that - but its got my blessing fwiw [21:13] wolsen, ack & thank you sir [21:13] beisner, thank you! [21:14] * wolsen turns towards ackk's proposal then to skylerberg's [21:16] anyone able to help me diagnose my bundle parse errors ? [21:27] nevermind, I needed to use “juju quickstart” instead of “juju-deployer" [21:27] firl: huh, quickstart and deployer should both work, what were teh errors? [21:28] http://pastebin.com/vds8YYDh [21:28] http://pastebin.com/w9UckkRi [21:28] nevermind the quickstart just failed more gracefully [21:28] haha [21:28] firl: heh :D [21:28] what version of deployer do you have installed? [21:30] it seems to be an issue with parsing constraints [21:30] yeah, worked on my old bundle file [21:30] what changed in this bundle file? [21:31] switching from v3 to v4? I think? I mimic’ed the openstack-base charm [21:31] and then added in my constraints [21:31] insteresting [21:31] let me check something [21:31] and the quick start might have worked? still waiting for the gui to come up [21:32] firl: it's possible [21:32] it seems that "cannon unmarshal string into Go value type []string" means that the API is expecting a list of strings and not just a string [21:32] firl: lets wait for quick start to succeed or fail [21:32] and go from there, I have some ideas [21:33] cool [21:33] https://api.jujucharms.com/charmstore/v4/bundle/openstack-base-36/archive/bundle.yaml [21:33] that constraints only has 1 as well [21:34] firl: yeah, I've not used the v4 bundle format as much as I should [21:34] or maybe it’s thinking tags=[“tag1”,”tag2”] [21:34] I don't think so, but it's possible [21:34] rick_h_: ^ [21:40] juju gui timing out to install, I will blow away and re bootstrap with dragging the yaml file to the gui. I don’t think the gui has a place for maas tags though [21:40] firl: it may not expose it explicitly, but it should respect it [21:41] cool [21:43] firl: for what it's worth juju will soon be gaining the ability to just do `juju deploy ` so it'll be a lot more cohesive and we can finally drop things like deployer and quickstart for bundles [21:43] that's why we're transitioning to the v4 bundle format [21:43] sweet; yeah I’ve really enjoyed everything I have encountered when it comes to juju [21:44] bringing a co worker with me to the summit too [21:46] firl: \o/ aweseome, see you there [21:46] Hi, I'm having some trouble deploying juju-gui to my bootstrapped environment. I run "juju deploy --to 0 --repository=/srv/charmstore local:trusty/juju-gui" and then "juju expose juju-gui", and for a while the service-status is maintenance and the agent-status is running the install hook and the start hook, and then eventually the service-status changes to 'unknown' and the agent-status changes to 'idle', and even though the agent-state is started, going to the [21:47] ntpttr: hey, so since 1.24 juju introduced a new extended status. a service status of unknown and agent-state of idle is for charms not yet using extended status. So that means the juju-gui should be deployed [21:47] ntpttr: are you having issues loading it in your browser? [21:47] marcoceppi: Yeah, the browser is showing a spinning wheel under "Connecting to the Juju environment" [21:48] ntpttr: hum, what provider are you using? local, maas, openstack? [21:48] marcoceppi: maas [21:48] marcoceppi: I have a corporate proxy I'm working behind, but I did configure that in environments.yaml [21:49] ntpttr: ah, that might be the issue [21:49] ntpttr: so the juju-gui uses websockets to communicate with the juju environment [21:49] ntpttr: what browser are you using? [21:49] marcoceppi: Firefox [21:50] ntpttr: if you open the developers tool console in firefox and hard refresh the juju-gui I imagine you'll probably see a connection error pop up [21:51] marcoceppi: Yeah it's saying can't establish a connection to the server at wss://node0vm0.maas/ws. It was working yesterday for me, but then I updated my juju by adding the stable ppa [21:51] ntpttr: so what version were you on before? 1.22? [21:51] marcoceppi: It's also saying the connection was interrupted while the page was loading [21:52] marcoceppi: I'm honestly not sure, whichever one comes with the orange box build [21:52] ntpttr: are you doing this on an orangebox right now? [21:52] marcoceppi: yes [21:53] ntpttr: what does `env | grep "proxy"` produce on the machine you're trying to access the GUI from [21:54] marcoceppi: It produces all of my proxy urls for http_proxy https_proxy ftp_proxy and no_proxy. Yesterday I was having a connection issue as well and I fixed it by adding ".maas,10.14.4.1" to no_proxy [21:54] marcoceppi: Is there an easy way to downgrade juju to what I had before? [21:54] yeah, you're going to have to do that again [21:54] ntpttr: you could remove the stable ppa [21:55] marcoceppi: drag to juju-gui seems to work so far. [21:55] marcoceppi: and after I do that if I update it will downgrade? [21:56] ntpttr: you'll have to either apt-get remove juju-core and re-install or do `apt-get install juju-core=1.XX*` whatever the version is [21:56] ntpttr: I suggest the former [21:56] firl: nice! glad to hear the drag and drop is working so far