=== anthonyf is now known as Guest1386 [08:14] hey, is there a base vivid charm for running some bleeding edge version of vivid? [08:14] or if not, is there a way one can easily make one [08:15] thanks! [08:15] it's just for testing that some of the software we're building runs on daily vivid [08:28] good morning o/ [08:31] guys, anyone can give me some help with the nfs charm? [08:33] simply, I have my charm, and I'm configuring load balancing on it... I figured that I need to "share" many folders on my charm, and wanted to use the NFS charm to accomplish that [08:33] with "share" I mean all the units of the charm should have the same file system [08:34] does the NFS charm work as a NFS server? [09:15] zyga: there's no real difference except that you can't use upstart, you have to use systemd for init jobs. Otherwise just create a vivid directory in your local charm repository and start creating [09:15] Muntaner: the nfs charm does work as an nfs server [09:18] marcoceppi: hmm, I'm not that familar with juju, could you give me an example? [09:20] jamespage, do you have a second for https://code.launchpad.net/~gnuoy/charm-helpers/nrpe-proxy/+merge/253166 ? [09:20] zyga: depends on what you're trying to test, basically do this: http://paste.ubuntu.com/10614347/ [09:21] doh, forgot to include how to deploy [09:22] zyga: http://paste.ubuntu.com/10614351/ [09:25] marcoceppi: looking [09:25] marcoceppi: we want mostly-vanilla vivid to run our tests on top [09:25] marcoceppi: so just using juju as a tool to get to instances and discard them later [09:25] marcoceppi: thanks, I'll give that a try :-) [09:26] zyga: `juju charm get ubuntu` will get you the Ubuntu charm which does absolutely nothing (but get you an instance of the series you deployed it under) [09:26] you'll need to make sure you have charm-tools installed from ppa:juju/stable for that to work though [09:29] marcoceppi, I'm following your wordpress charm, but now my "mount" command just hangs [09:29] Muntaner: that's odd, do you have any logs? [09:30] it's rather late my time currently, I'm about to log off for the evening, but that's where I'd start [09:30] marcoceppi, nothing more than the juju logs [09:30] Muntaner: syslog and dmesg would be helpful for mount issues [09:30] Muntaner: are you testing on the local provider? [09:30] marcoceppi, no, it isn't local [09:31] Muntaner: hum, if you post logs for syslog and dmesg I can take a look later on today when I get back online === CyberJacob is now known as zz_CyberJacob [09:32] marcoceppi, thanks, I'll do it asap [09:36] marcoceppi, can't find anything special in the logs, can paste them now if you want [10:05] guys, anyone of you has used nfs charm? [10:05] mount hangs and can't understand why [10:25] Muntaner: I haven't recently - which charms are you using it with? [10:28] lazyPower, my personal charm. Now I discovered that probably the nfs charm in the store isn't working - I can see in it that there are no nfs daemons running [10:28] lazyPower, by creating my "exports" manually, seems like I'm "adjusting" it [10:28] but can't figure why it isn't working [10:28] Muntaner: are you running the NFS charm on the local provider? [10:28] lazyPower, no, I got it by the store: my personal charm is ran in local [10:29] Muntaner: i mean using the local provider to deploy the nfs charm [10:29] which it sounds liek you are - and that will not work. the Militant App Armor profiles assigned to lXC containers will prevent it [10:29] Muntaner: if you look @ the readme for NFS - it tells you how to work around this [10:29] https://jujucharms.com/nfs/trusty/0 - under the heading "On the LXC host" [10:29] lazyPower, no: I deployed it from the store. My cloud is a private openstack installed in a all-in-one server [10:29] not using lxc [10:30] ah ok, i misunderstood [10:30] np :) [10:30] let me bootstrap AWS and kick off a NFS deployment, 1 sec [10:30] thanks lazyPower [10:37] lazyPower, I figured that the NFS charm hasn't hooks like the "config-changed" or other... I'm a bit confused [10:42] Muntaner: its not a very well formed charm - missing the config-changed hook. It's got immutable configuration in its current form [10:44] Muntaner: can you file a bug against the charm here https://bugs.launchpad.net/charms/+source/nfs and attach any relevant logs? [10:46] lazyPower, and how does it work with wordpress? lol [10:46] btw ok lazyPower, will do it asap :) [10:46] Muntaner: it just came online for me - starting to probe the env to see whats running [10:49] ok lazyPower fine [10:49] when I deployed it, I got a machine with no daemons running [10:50] did you add the nfs relation to the charm? [10:50] looking at the hooks - it doesn't do anything until the nfs-relation-changed hook is fired, zero exports = no daemon can start. [10:52] yes lazyPower, I did it [10:54] but nothing gets wrote in exports... [10:54] and so, no daemon is running [10:54] wp & mysql is coming up, i'll have a result shortly [10:55] lazyPower, I never tested it with WP, to be honest: I always did tests with my personal charm [10:57] Muntaner: confirmed - missing exports here as well. [10:57] ping back with the bug number and i'll triage it accordingly [10:58] lazyPower, so it basically never worked? [10:58] it did at one time or it wouldn't be in the store, unfortunately its got a bug filed against it as an unmaintained charm - which means its exhibited problems in the past and was at risk of removal [10:58] this is only further compounding that problem, that its utterly broken with nobody actively maintaining the charm [11:00] Difference between deploying bundle with quickstart or juju-deployer? | http://askubuntu.com/q/597826 [11:00] lazyPower, I'm writing the bug [11:06] lazyPower, https://bugs.launchpad.net/charms/+source/nfs/+bug/1433036 [11:06] Bug #1433036: NFS charm does not create "exports", daemons not running [11:13] lazyPower, are you actually awake and working, isn't it early for you? [11:13] lazyPower, ^^ I have a quick question if you have 1 minute [11:14] mattyw: yup [11:15] * lazyPower reads backscroll [11:15] mattyw: i dont see anything in the backscroll - fire when ready === kissiel is now known as kissiel-lunch === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kissiel-lunch is now known as kissiel === kadams54 is now known as kadams54-away [14:38] hey ho o/ === urulama is now known as urulama|afk [15:11] hey guys [15:11] jamespage: You around? [15:11] how do I create a bundle? the official page seems to be dead [15:11] -> https://jujucharms.com/docs/charms-bundles/ [15:14] Muntaner: apologies there, bug that we've got a fix landed and working on a release today: http://qa.storefront.theblues.io:6543/docs/1.20/charms-bundles is the page on our QA site atm [15:14] Muntaner: pulled from the juju docs at https://github.com/juju/docs/blob/master/src/en/charms-bundles.md [15:16] rick_h_, thanks! [15:16] Muntaner: np, let me know if you have any questions === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [15:44] rick_h_, where should I put the local charms in order to be deployed in a bundle? [15:45] Muntaner: bah caught me. Makyo do you know much of the deployer/local repo setup? I know you can define a local repo do you need the whole series subdirs? [15:47] rick_h_, Muntaner a quick glance shows that yes, you do need series subdirs, but I don't know too much beyond that, sorry. I haven't had to touch the charm stuff yet. [15:48] Makyo, no problem, it isn't urgent. [15:48] Makyo, so I can't deploy a bundle with charms on my filesystem, basically? [15:49] Muntaner: yes, with the deployer [15:50] rick_h_, my bundle has mysql and haproxy took from the official store, and two local charms. As far as I can see, the bundle fails since it can't find my local charms [15:50] dunno what to do exactly [15:50] Muntaner, AFAICS, you can, with a charms// directory structure and the bundles in charms dir. The deployer makes a few assumptions around there. === kadams54 is now known as kadams54-away [15:52] Muntaner, eg: deployer doesn't obey --repository flags, so it's PWD sensitive [15:52] Makyo, I'm trying, but get a "An error occurred while deploying the bundle: ('No charm metadata @ %s', 'trusty/...blabla" [15:52] Makyo, simply trying by drag and dropping the bundle file into the juju-gui, not CLI [15:53] Muntaner, oh, hm, yeah I don't know that that's supported yet, unfortunately. [15:55] you can use the format "charm: local:precise/memcached" in your bundles, just make sure JUJU_REPOSITORY is set [15:55] juju-deployer will handle that, anyway [15:55] Muntaner: the gui doesn't support local charms since it can't upload them to juju for you [15:56] Muntaner: so you have to use the deployer which uses local juju commands to get the charm deployed [15:56] Muntaner: ^ as tvansteenburgh says [15:56] Muntaner: the GUI only supports charms that are in the charmstore as it can then pull them down from the public store api [15:56] rick_h_, I never used the deployer, I always deployed local charms via juju deploy [15:58] Muntaner: yea, time to use the deployer if you want a bundle with local charms from your hard drive [15:58] rick_h_, will try immediately :) [16:03] rick_h_, I'm getting a "deployer.cli: Config files must be specified" [16:05] Muntaner: you need to pass -c bundles.yaml I believe [16:05] Muntaner: that points to your bundle file [16:07] rick_h_, great! seems to work \o/ [16:08] Muntaner: woot! [16:16] rick_h_: do you know who owns the neutron-api charm? [16:17] apuimedo: looking at https://jujucharms.com/neutron-api/trusty/8 it's openstack-charmers with commit history coming from liam and james mostly (right column 10 revisions) [16:17] apuimedo: the openstack-chamers own it, it's a team of people maintaining the openstack charms [16:18] rick_h_: marcoceppi: I wanted to ask about neutron-api support for midonet. I was talking about it a bit this morning with lazyPower as well. [16:19] as I told lazyPower, for it to be clean/ideal it should be patches to neutron-api that enabled it to add a relation to midonet-api with interface midonet (it would get the IP of that service from that) [16:20] the second less ideal way, would be to do like the nsx and add config options and make the deployers check the IP of the juju deployed midonet-api [16:20] and set it to neutron-api [16:20] apuimedo: so you're looking to submit some updates to http://bazaar.launchpad.net/~openstack-charmers/charms/trusty/neutron-api/trunk/files ? [16:20] rick_h_: looking into the best fit for doing so [16:21] marcoceppi: email the juju list about the charms and possible ways of updating perhaps? [16:21] lazyPower also mentioned about having a subordinate charm that provided extra templates, but I'm not entirely sure how one would go about that [16:22] apuimedo: gotcha, so you're looking to chat out the best way to look into adding support. I'd probably suggest emailing the juju list and bring up the thoughts and let folks do some suggesting. [16:22] apuimedo: and the folks that run that charm will hopefully chime in there and you'll get some good feedback [16:23] rick_h_: ok. I generally prefer the short loop of IRC, but if the folks are not around, better send it to the mailing list. [16:23] apuimedo: yea, async sometimes is best [16:23] indeed [16:33] if a charm is in error state and you upgrade it, it will try to do it? or first has to be changed to "OK" it's state? [16:43] apuimedo: if you want to submit changes to openstack charms, it's done a certain way [16:44] apuimedo: https://wiki.ubuntu.com/ServerTeam/OpenStackCharms [16:44] thanks [16:44] apuimedo: you want to target the /next branch for whatever service for your merges [16:44] ok, that means that any change would not be ready until kilo [16:45] apuimedo: unless you need it to be merged ahead of time [16:45] apuimedo: you can signal in the merge that it should also land in /trunk [16:45] it needs to be in next, but changes can be ported to current /trunk [16:45] but they would be merged to next in any case, or do you do backports? [16:45] aha, I see [16:45] apuimedo: exactly [16:45] jamespage: it'd be great if that link was in the README for the charms under # Contributing as there's been a lot of questions around that wiki page [16:52] marcoceppi: how are the charmhelpers managed? I see that the charmhelpers installed with pip differ from those embedded in neutron-api/next [16:52] so if I want to add to charmhelpers/contrib/openstack, where would be the right place to go? [16:59] marcoceppi: nevermind, found it. with the lp:charm-helpers and then make-sync [17:01] hey asanjar or kwmonroe [17:02] do we have an example of a charm that let's you swap out OpenJDK/Oracle Java? [17:02] I'm putting together a list of examples. [17:03] something with a config option perhaps? [17:05] I have mysql/0 charm as "hook failed: "config-changed"", everything seems to be ok (the problem was a communication problem with the bootstrap), so I ran the "juju resolved --retry mysql/0" command, and then, the unit download some stuff from ubuntu.com and when it ended, gives this error: " No such file or directory: '/var/lib/mysql/mysql.passwd'" [17:05] that file doesn't exist and I don't know what to do.. any suggestions? [17:13] jcastro: cory_fu pulled the java logic into a java_installer script that does the switch based on arch: http://bazaar.launchpad.net/~bigdata-dev/bigdata-data/trunk/view/head:/common/noarch/java-installer.sh [17:13] kwmonroe, ok so like this isn't in charm helpers or anything? [17:16] jcastro: We went for a more pluggable implementation. So, the charm loads the script that kwmonroe linked using jujuresources and calls the script to actually install whatever Java it decides to. This allows a vendor to change how Java is installed just by changing a single line in the resources.yaml [17:16] perfect, thanks! [17:17] jcastro: Previously, we had special logic to install openjdk or use a resources.yaml entry to download a binary installer. Which was used depended on the cpu_arch [17:19] jcastro: With the new approach, the installer script(s) could be bundled into the charm under a resources/ dir, and the resources.yaml entries could choose the correct one with a file: URL [17:19] cucumber, I wonder if touching that file and rerunning it will help? [17:19] yeah jcastro, this is how we used to do it in charmhelpers if interested (line 149-168): http://bazaar.launchpad.net/~bigdata-dev/charm-helpers/framework/view/387/charmhelpers/contrib/bigdata/handlers/apache.py#L149 [17:20] this is great, any other reusable chunks while I am writing this up? [17:20] jcastro: kwmonroe's last link is the old way in charmhelpers. If you want the new way, we can show you that code (which is also in the bigdata charmhelpers branch) [17:21] when do you suppose this will land in proper charm helpers? [17:21] jcastro: we're working on a DistConfig reusable module. it lets you store immutable config in a dist.yaml: http://bazaar.launchpad.net/~bigdata-dev/charm-helpers/distconfig/revision/386 [17:22] ack [17:22] that's stuff you may want to configure (like users and groups), but don't want to change post deployment [17:22] It would be nice to get it cleaned up and ready during the sprint next week. The main difficulty is that we're still iterating on the big data charms to get them ready for the store, so there's a lot of churn [17:22] right, for now I just want a link to point to people who might be interested [17:22] out of curiosity did anyone show this to the ~ibm-charmers? [17:23] Not that I'm aware of [17:24] CH code using the installer script: http://bazaar.launchpad.net/~bigdata-dev/charm-helpers/framework/view/head:/charmhelpers/contrib/bigdata/handlers/apache.py#L159 [17:25] jcastro: i showed basic uname switching logic at the joint hangout a couple months back. iirc, i showed the cassandra special config for ppc64le: https://github.com/juju-solutions/cassandra/blob/master/hooks/cassandra-common#L365 [17:26] but i haven't shown them jdk switching [17:26] kwmonroe, can you take a note to post the JDK switching to the list at some point? [17:26] it might be useful to arm people as well [17:26] noted jcastro [17:40] Landscape openstack juju management | http://askubuntu.com/q/597979 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === anthonyf is now known as Guest85523 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === ericsnow_ is now known as ericsnow === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away [21:54] Does anyone on here know anything about ceph charms? [22:42] bdx: don't know about others, but I don't, sorry [22:42] better some response is better than tumbleweeds [22:44] bdx: we're using storage-ceph with the block-storage-broker [22:44] bdx: https://jujucharms.com/u/tribaal/storage-ceph/trusty/6 === kadams54-away is now known as kadams54 [22:51] bdx, some of the ceph devs and maintainers are in the UK time zone so you may want to mail the list to get some coverage there [22:52] bdx, where list == https://lists.ubuntu.com/mailman/listinfo/juju === kadams54 is now known as kadams54-away [23:11] arosales, thumper, blr: Thnaks [23:11] bdx, np, looking forward to seeing what your question about ceph is on the list :-) [23:14] arosales: My issue now stems from maas.....do you know if there a was to have maas provision(commission) a node whos only connected interface is not eth0? [23:15] It seems if there is not eth0 is connected the node fails commissioning and/or booting [23:16] I'v posted about this in #maas .... bit of a rough spot for me right now ;'( [23:17] * arosales was just going to suggest #maas . . . [23:18] arosales: Do you see my question in #maas? [23:18] I think this might be one for the mailing lists [23:18] bdx currently are you manually commissioning via mac? [23:19] * arosales just joined #maas so I missed the backscroll [23:19] arosales: No. I have all nodes pxe and ipmi managed by maas [23:20] I ~think~ you could manually bring this in, but I am sure your are looking for programmatic way to do this, in case you have a few of these type of machines [23:20] Entirely [23:21] ya [23:21] and it pxe boots ok, independent of maas, correct? [23:21] yea. [23:22] All nodes pxe boot and commission great when eth0 is connected to my maas-mgmt net [23:23] If I disco eth0 and connect eth{1,2,3,4,5} and try to commission/boot the node it fails and hangs right after ipmi power on (ipmi has a dedicated interface) [23:25] The reason behind this is I need to have ceph-public-network == maas-mgmt net so nucs that host services like ceph-mon and ceph-radosgw can communicate on ceph-public-network(10G) [23:26] ya, I could also see this where folks have mgmt over eth0, but have a high speed nic for all other traffice. [23:27] Totally...thats kindof what I have going on... [23:27] bdx, atm unfortunately I think the list may get you the most eyes [23:27] juju list as well as maas mail list [23:27] Totally. Thanks again arosales. [23:28] I can see what you are trying to do, but don't have a readily available answer :-/ [23:28] bdx, I'll look for the post and encourage others when they are back on-line to reply :-) [23:29] arosales: Thanks sooo much.....I have been trying to get this stack up for a while now....every bit helps. ThankS! [23:29] bdx, oh np at all, just wish I had a better answer for right now [23:29] bdx, but post to the list and I'll be sure to chase it down. [23:30] bdx, in that post be sure to mention what mass version you are using [23:31] Entirely. [23:31] thanks, I think this will be very helpful to others too === kadams54-away is now known as kadams54