[02:38] firl yeah :( [04:09] lazypower|travel: gotcha, is there a benefit to doing it this way over having k8s bootstrap the env itself? [04:09] The flannel networking is a bolt on for cross host communication. The charm itself does deliver a k8s bootstrapped cluster [04:10] we're leveraging hyperkube to standup the cluster, so we're closer in alignment with their reference arch [04:11] firl - unless i completely missed the questions intent.... [04:11] http://kubernetes.io/v1.1/docs/getting-started-guides/juju.html [04:11] which is possible, its 5am [04:11] hah yeah gotta love time zones [04:12] firl - ah yeah, dont trust that document :) thats in reference to our older charm work where we were delivering k8's entirely via CM [04:12] hah kk [04:12] it was klunky to say the least, but it was an important evolutionary step in our understanding of their stack [04:12] thats actually a prime topic in my FOSDEM track today [04:12] gotcha [04:12] too bad I couldn’t come haha [04:13] So I should wait until monday? will it get pushed to the charm store [04:13] it'll be recorded, and since we're buds, i'll do a special after-session for ya if you're interested :) [04:13] or just fixed so I can deploy the bundle with charm build layer [04:13] It’d be cool to get a better understanding [04:13] I am going through bulletproofmeteor and they are recommending kubernetes [04:13] yeah, what I'll do is after FOSDEM lets out and I get settled in Gent i'll dive back into fixing that stilly bug. [04:13] and since the meteor charm is broke, I figured why not [04:14] i can ping you here or via email once i've got a fix proposed, or you can star the layer-k8s repo [04:14] as I gate all that work through mbruzek [04:14] celpa.firl@gmail.com [04:14] ( I will star the github also ) [04:14] awesome. I'll reach out when i've got the fix in place for the flannel containers :) [04:14] I have a build node all set aside for it [04:14] just for good measure [04:15] I can at least smoke test it for ya [04:15] since I replicated it [04:15] <3 really appreciate that [04:15] so I should just do a single docker node for now [04:15] sure, actually if you remove 'layer:k8s' from the layer.yaml [04:15] rebuild and deploy with a single k8s node + single etcd service [04:16] you should be in like flynn [04:16] sorry, remove 'layer:flannel' [04:16] and you can even scale that, you just wont get distributed pods working. the scheduler is intelligent enough to know this in my testing [04:16] lol gotcha I was like … yeah it’s not there [04:16] ok [04:16] I will do that for now and see how it goes [04:16] then once it’s ready I can add the overlay networking [04:16] yep! [04:17] in-place upgrade should take care of that missing component [04:17] any DTR charms you recommend [04:17] whats DTR stand for again? [04:17] docker trusted registry [04:17] oh! [04:17] ( to use outside of kuber ) [04:17] for it to reference [04:18] I started on, but haven't finished it yet, we spiked on TLS support instead of the registry [04:18] gotcha [04:18] with that being said, I can see hvaing a docker distribution charm in the next 30 days or so [04:18] if you're feeling spunky, i can get you early access there as well [04:18] yeah, I don’t mind testing/contributing [04:19] all of my charms currently are private only repo’s [04:19] I also haven’t written anything for 1.25.3 [04:19] so status/update/blocking is still a foreign concept [04:19] well, 2.0-alpha1 is coming really soon [04:19] ( for dev that is ) [04:19] so be prepared for nomenclature changes, and even more awesome [04:19] haha [04:19] yeah running trusty-liberty now [04:20] as in single controller - many models (thats right, one state server for many environments) [04:20] nice [04:20] and thats just the tip of the iceberg :D [04:21] ya, i’m a fan of it [04:21] have fun at the FOSDEM, I will try to deploy the bundle without flannel now [04:22] firl - thanks :) feel free to reach out if you encounter any further issues. I'm happy to lend a hand. And I really appreciate the testing/patience while we iron out our next major release of the charms [04:23] hah yeah, it’s been a bumpy ride. It’s nice though to have configuration around the environments in a private cloud [09:45] firl, so there was a maas 1.9 bonding + juju bug but it should be fixed in .3 [10:48] yoyoyo [10:48] ghent-charmers: what hotel are we in? [11:31] is it possible to set a charm config from within an action? [21:08] lazypower|travel: if you messaged me I missed it [21:08] I saw a notification but the scroll history buffer is not big enough to show me what it was [22:12] firl - nope but here's a repost - nagyz firl, so there was a maas 1.9 bonding + juju bug but it should be fixed in .3 [22:12] hah, k [22:12] lazypower|travel: I am still on for tomorrow for setting up the cluster let me know when you want me to test it [22:13] If I can get it, will deploy over maas and openstack [22:13] firl - will do. I'm still preoccupied with prep for the charmer summit, but for sure I'll be hacking on that regression [22:14] looking like its going to be first up tomorrow morning unless i get completely shanghaid [22:14] no worries, I can do tuesday too if that’s easier. no worries on my end [22:14] Ok, lets shoot for that so i have some breathing room :) [22:14] * lazypower|travel adjusts the todo due-date [22:15] haha k [22:15] just realized I was on the http://summit.juju.solutions/ website lol [22:17] :) Yep! [22:17] * lazypower|travel chants "one of us, one of us" [22:18] haha [22:29] lazypower|travel, yeah but it's not :( [22:29] lazypower|travel, I've been trying with .3 yesterday as well [22:32] nagyz at least you have something to track though [22:32] should I reopen the bug? [22:34] nagyz - link to the bug? === lazypower|travel is now known as lazypower|summit [22:52] I think this is the one that firl mentioned: https://bugs.launchpad.net/juju-core/+bug/1516891 [22:52] Bug #1516891: juju 1.25 misconfigures juju-br0 when using MAAS 1.9 bonded interface

[22:53] but my bug is that SOMETIMES it has some weird error. [22:53] 50/50 :) [22:55] :-/ hrm [22:55] yeah, if you have logs/data to back up the hinky behavior, it'll go a long way to re-open the bug with logs [22:56] I've posted the pastebin yesterday - do you have logs or shall I relink them? [23:01] nagyz - same output contained in comment #19? https://bugs.launchpad.net/juju-core/+bug/1516891/comments/19 [23:01] Bug #1516891: juju 1.25 misconfigures juju-br0 when using MAAS 1.9 bonded interface

[23:02] I don't run VLANs [23:05] ok, yeah - i would most def attach the logs and at bare minimum poke the bug [23:07] the thing is it's not that it misconfigures the interface [23:07] it's just.. something. I have no idea what. I can ssh to the node and whatnot.