[00:27] veebers: extra data point - it looks like it might also happen when new controller machines are added? It looks like the mysql unit responds to the config-changed, and updates it status to "maintenance", at which point the mediawiki unit goes to "blocked", and then the mediawiki unit never changes back to "active" [00:59] babbageclunk: cheers, I think I know when it's happening, I just have to confirm. (the db-relation-departed hook removes the canary file it uses to say "I have a connection") I think it's being triggered and miss-handled [01:00] veebers: ah, ok [01:01] babbageclunk: do u have a min...to discuss upon this issue i mentioned this morning [01:02] ? [01:02] vino: sure [01:02] hangout ? [01:03] I'm in standup [01:10] Oh, babbageclunk btw: juju show-status-log mediawiki/0 [01:22] veebers: oh right - I always forget that command. Will that show all the hooks that get run? [01:23] babbageclunk: aye it does, in the order they where run, with the message etc. v/ helpful for what I want [01:23] cool === jhebden is now known as jhebden-afk [02:07] Hello! Is there anyone out there with knowledge of using a pre-existing charm as a layer if it does not exist in https://github.com/juju/layer-index/tree/master/layers ? [02:09] My team is running into this: https://bugs.launchpad.net/juju/+bug/1764317 [02:09] Bug #1764317: bionic LXD containers on bionic hosts get incorrect /etc/resolve.conf files

[02:11] Our idea of a fix is to dynamically generate the DNS data using a charm hook and merge it into the affected preexisting charms, but we are not sure of the best way to get this done. === jhebden-afk is now known as jhebden [03:02] anastasiamac: is there a way to check logs on status of upgrade at controller side ? [03:02] veebers: if anyone can help me on the above question.. [03:03] vino: u can check controller logs: [03:03] swicth to controller model, ssh inot machine and broswe the logs [03:06] Is there a why to find the gh/lp page for a charm? (i.e. https://jujucharms.com/mediawiki/ I want to propose a fix0 [03:12] veebers: generally there's a link on the charmstore page [03:13] ah right, I had thought so too. Perhaps I'll try file a bug with a pastebin of the diff to fix it. [03:13] veebers: https://bugs.launchpad.net/charms/+source/mediawiki [03:13] babbageclunk: cheers L:-) [03:32] where can one see the current bugs for 2.4? I can only see fix committed or in progress ones. [03:42] veebers: for next beta - https://launchpad.net/juju/+milestone/2.4-beta2 [03:43] veebers: or all existing - https://bugs.launchpad.net/juju [03:44] anastasiamac: ah right, I had assumed there would be bugs targetted against 2.4 in general [03:45] veebers: so have i but it does not think so.. it looks like tags were used at one stage but all tagged 2.4 bugs have been addressed :) [03:46] anastasiamac: that's good, seems we're on top of the fixes going into 2.4 [03:46] anastasiamac: is https://bugs.launchpad.net/juju/+bug/1465317 still valid? [03:46] Bug #1465317: ubuntu devel macos win: panic: osVersion reported an error: Could not determine series

Fix>

[03:48] veebers: i do not have more info than on the bug.. mayb? :) [03:49] veebers: fwiw it hink we've had couple of attempts at resolving it for good (as opposed to band-aid solutions) but obviously have not resolved it.... would b brilliant to fix :) [07:49] Hello! We are trying to make use of "ansible" with juju. How would we make best use of that without needing to rebuild existing charms for our use-case? I've seen a "layer-ansible" somewhere but I would hope to get some input on the subject. [08:08] hi eriklonroth from what I've read the ansible layer should fill that need. Do you want a standalone charm or do you want things on top of the ansible layer? [08:10] Thanx! I think I need something standalone as I'm not yet sure how to puzzle this into the different services we are to create. It would be good to not have to go through a vast testing phase each time a change is made to ansible layers or/and the target charms. [08:10] ... a subordinate charm is perhaps the way to go ? [08:12] eriklonroth: I think it depends, in your case I would try first with a normal charm and https://github.com/chuckbutler/ansible-base layer to test playbooks. Once that works it might be easier to determine what you want the charm to do :) [08:13] https://github.com/chuckbutler/charms.ansible 's second method might be interesting to look at too [08:13] Thanx for that advice. What are your thoughts about this thing I just googled. v [08:13] https://github.com/cmars/juju-ansible [08:38] eriklonroth: I think it's outdated honestly, if it works sure nice thing but I don't see a reason to not use the reactive framework and the existing layers. It all depends however what you want to do with ansible and on what level [08:39] I see [08:39] OK, I'll take your advice to begin with and see where we end up. Do you know of any examples I can look at ? [08:45] unfortunately not really, haven't used it myself maybe others here can help [09:10] hi, when I run "juju deploy --constraints tags=compute --config compute.yaml nova-compute" and check the status the public address is by default coming from the IP of of the first NIC in MAAS to that node [09:12] https://paste.ubuntu.com/p/wGQJnxDMdm/ [09:13] Its always taking the IP from storage space while I expect from the management space. Is there a way to specify that in compute.yaml? [09:19] srihas: You probably want to look at bundles https://jujucharms.com/docs/2.3/charms-bundles#binding-endpoints-within-a-bundle or specifying binds via the cli https://jujucharms.com/docs/stable/charms-deploying-advanced#deploying-to-network-spaces [09:22] zeestrat: aha, will see into it. seems I understood something thatnk you [09:23] zeestrat: but if there are 4 subnets in "management" space which IP will come to my_ip in the nova.config? [09:23] when I bind the "management" space [09:26] srihas: That's a good question. Not sure. Does the machine in maas have interfaces for all those subnets? [09:26] zeestrat: no, I have given static IP from one of those subnets specifically [09:26] I hope it solves [09:28] srihas: Gotcha. Hope it does the right thing. [12:09] zeestrat: no, it didn't it simply ignored the "bind" :( [12:33] yayy, it worked, thanks zeestrat [13:02] is there a way to tell juju to install ocata version? I got Mitaka version of nova-compute. Can someone help? [13:20] srihas: You can use the openstack-origin option for most of the openstack charms, for exmaple "openstack-origin: cloud:xenial-ocata". [13:21] srihas: If you look at the main openstack bundle from the openstack-charmers (https://jujucharms.com/openstack-base/, see bundle.yaml) you can see the "openstack-origin" option used for the latest release, queens. [13:24] zeestrat: have you used the mysql-shared or pgsql interface recently in a charm (reactive) by any chance? [13:25] zeestrat: /win35 [13:25] nvm [13:27] TheAbsentOne: No I haven't, sorry. [13:29] never lucky x) [13:32] zeestrat: I think I understood it again, thank you [13:39] https://pastebin.com/4znUZhvv confused why this is not enough for the mysql-shared interface, got things working on mysql interface but not on shared where I can share dbname-credentials :/ [16:18] TheAbsentOne: it helps if you post the github repo with all of the charm code, off the top of my head I see a slew of issues there [16:20] TheAbsentOne: here's how I've done it in the past https://github.com/omnivector-solutions/layer-documize/blob/master/reactive/documize.py [16:22] TheAbsentOne: ^ code is now pretty stale though, it should be implemented using reactive Endpoints, and flags etc etc [16:23] I'll ping you with an update when I've refactored that to use the new bits [16:23] TheAbsentOne: hopefully that will get you thinking in the right direction though [17:17] TheAbsentOne: it helps if you post the github repo with all of the charm code, off the top of my head I see a slew of issues there [17:17] TheAbsentOne: here's how I've done it in the past https://github.com/omnivector-solutions/layer-documize/blob/master/reactive/documize.py [17:17] TheAbsentOne: ^ code is now pretty stale though, it should be implemented using reactive Endpoints, and flags etc etc [17:24] thanks bdx, gonna take a look. I might get back to you if I find myself struggling once again; great link! [17:24] np [17:57] bdx: it seems I used the configure function wrong, but weirdly enough the flag on database.available is never set. There must be another issue [17:58] TheAbsentOne: make a very minimal example charm to zero in and debug just the mysql relation bits [17:59] you could do it with two handlers [17:59] let me see if I can get a quick example for you [18:01] bdx: I might have found it but my battery is dying on me so have to go. If I'm still in need of help I'll ping you for some help after putting everything online too, thanks for the guidance though! [18:01] np === hml_ is now known as hml [20:17] Morning team \o/ [20:58] I'm pretty sure I've seen chatter re: percona-cluster and hooks issues recently, can anyone update me? [21:53] To use a local copy of a reactive charm I need to build it right? I recall doing something like this *ages* ago but the details escape me [21:57] heh, I think I have it. 'charm build' has done something without error, going to try that z^_^_ [22:02] TheAbsentOne: https://github.com/jamesbeedy/mysql-shared-example [22:04] damn bdx I was just typing I got it working xD [22:05] :) [22:05] it was good for me too [22:05] there is only one thing fishy about my status's/flags when the database already exists [22:06] but in all honesty, we need more examples of this bdx like real "state of the art examples" [22:07] TheAbsentOne: you can look at the code of any of the upstream charms <- usually what I do when I'm in need of examples [22:07] I pretty much have the same now of your minimal example xD sorry for the work you had to do [22:07] and well as a matter of fact I was looking at https://github.com/omnivector-solutions/interface-conda/blob/master/provides.py for some inspiration [22:08] I'm still having issues in creating my own interface [22:23] anyway I'm off to bed, thanks again for the help bdx! Goodnight [23:04] ugh, I've been watching the wrong application thinking it's the one causing the failure :-\ [23:05] veebers: just knowing that gets u closer to determining the root cause [23:06] anastasiamac: at teardown I'm seeing the machine being torn down but not the application and it keeps trying. That seems odd right? if it's machine is gone there essentially is no more application [23:06] kelvinliu: where you looking at percona-cluster hooks issues recently? [23:06] veebers: not necessarily [23:06] veebers: if it's a unit than yes, however, the applictaion is k to exist without any units (machines) [23:07] anastasiamac: oh ok, cool I'll continue digging [23:07] I'm wondering if it'll be like yesterday where I spent hours digging all to come up with a 3 line fix ^_^ [23:15] well, they do say that simplicity is ingenuios (or smth like that)