=== frankban is now known as frankban|afk [01:36] OK well that explains why I could use this nick earlier [04:21] =1 [07:03] how do I change location from which Juju downloads lxc images? === frankban|afk is now known as frankban [09:25] lazyPower: this RESTARTS column stress me a bit, but since I scaled them 1->5, at least they do not restart at the same time, so no interruption normally... do you have this kind of behavior in your labs? it's the same "spec.containers{dnsmasq} Warning Unhealthy Liveness probe failed: HTTP probe failed with statuscode: 503" error at everytime [09:25] http://paste.ubuntu.com/24012539/ [09:26] (oh, I'm not polite today : Hello!) === Guest49556 is now known as CyberJacob [10:00] im unable to view the charmpage of my unpublished charm in the new charmstore gui [10:00] i can see the charm in my list of charms, but when i click on the "View" button it says "There was a problem while loading the entity details. You could try searching for another charm or bundle or go back." === CyberJacob is now known as Guest40250 === Guest40250 is now known as CyberJacob === scuttlemonkey is now known as scuttle|afk [14:02] tvansteenburgh: ping [14:02] rick_h: yo [14:02] tvansteenburgh: hey, can I steal a few minutes to talk libjuju usage stuff sometime? [14:02] rick_h: yeah sure === kirkland` is now known as kirkland [15:23] Zic - no, my dns container has 4 restarts in 61 days [15:24] Zic - i'll shop it around though see what IS is seeing in their long running deployment [15:43] tvansteenburgh1: is this a not yet done method then? https://github.com/juju/python-libjuju/blob/master/juju/controller.py#L209 [15:45] rick_h: correct === tvansteenburgh1 is now known as tvansteenburgh [15:47] rick_h: if i know someone needs it i'll prioritize it, so you can either file a bug for that, or take a shot at implementing it and post a PR [15:47] tvansteenburgh: yea, just poking at the facade stuff and will see what I can do ty [15:56] lazyPower: we have 124 pods which is running (if it's tied to the number of pods) [15:57] Zic - well i only have 36 running, and only 1 dns pod [15:58] Zic - so your deployment volume is much higher. have you been watching the resource utilization of the pod in the dashboard? [15:58] is the DNS service using quite a bit of ram? any conntrack table issues in the logs? [15:58] Zic - i ask because rimasz found some issues with a competing solutions dns deployment, it filled up the conntrack table and started dropping packets like crazy before the pod itself was terminated due to failing health checks. [15:59] that *might* be a similar situation but i doubt it, i would think you'd have found other issues as a symptom if that were the case [16:25] lazyPower: I got some dnsmasq[1]: Maximum number of concurrent DNS queries reached (max: 150) logs that I success to collect directly via `docker logs` (kubectl logs on the kube-dns pod returned nothing) [16:26] it's maybe the problem [16:51] Hi *, i am trying to implement a python juju client, but when using the modle.deploy() i get an error: [16:51] Exception in thread Thread-6: [16:51] Traceback (most recent call last): [16:51] File "/usr/lib/python3.5/threading.py", line 914, in _bootstrap_inner [16:51] self.run() [16:51] File "/usr/lib/python3.5/threading.py", line 862, in run [16:52] in the doc i found [16:52] :param str to: Placement directive, e.g.:: [16:52] '23' - machine 23 [16:52] 'lxc:7' - new lxc container on machine 7 [16:52] '24/lxc/3' - lxc container 3 or machine 24 [16:52] If None, a new machine is provisioned. [16:52] but actually passing the simple machine number gives me the above error [16:52] can somebody give me a hint? [16:52] thanks in advance [16:52] tvansteenburgh: -^ [16:54] lorenzotomasini: can i see your code and the full traceback? [16:54] sure [16:55] so my code [16:55] async def deploy_local_charm(charm_dir_path, application_name, number_of_units, machine_number=None, model=None): [16:55] application = await model.deploy(charm_dir_path, [16:55] application_name, [16:55] num_units=number_of_units, [16:55] to=str(machine_number)) [16:55] log.debug("deployed application: %s", application) [16:55] and the full stack is: [16:55] Traceback (most recent call last): [16:55] File "/usr/lib/python3.5/threading.py", line 914, in _bootstrap_inner [16:55] self.run() [16:55] File "/usr/lib/python3.5/threading.py", line 862, in run [16:55] self._target(*self._args, **self._kwargs) [16:55] lorenzotomasini: please put it in a pastebin [16:56] ho ho ho [16:56] https://p.rrbone.net/paste/F3bTxZTB#Z1bY+UVIQ-t10wSxo5fHIpgaxppO58da3fgBGXUt2w/ [16:57] lorenzotomasini: okay, and what is the value of machine_number? [16:57] 112, int [16:58] what should it be? [16:59] actually it is a str [17:00] lorenzotomasini, tvansteenburgh: the docstring is dated, and it's my fault. Sorry :-/ You need to pass a list to "to". [17:00] So ['112'] should work. [17:01] petevg: i try this thanks [17:01] i think it also needs to be parsed [17:01] tvansteenburgh: you're right. [17:03] tvansteenburgh: sry my phyton is not really good, how should i parse it? [17:03] to=placement.parse(machine_number) [17:03] from juju import placement [17:03] lorenzotomasini: your Python isn't at fault. It's a tricky bit of code. (Thank you, tvansteenburgh) [17:03] something like that [17:04] ah but machine_number still has to be list of string or just str ? [17:04] just string, placement.parse will convert it [17:04] ok thanks [17:04] i'll let you know in some minuts [17:05] lorenzotomasini: example here https://github.com/juju/python-libjuju/blob/master/juju/application.py#L91 [17:05] tvansteenburgh: I'll submit an update to the docs shortly. [17:06] petevg: thanks! [17:06] np [17:08] found this [17:08] to=placement.parse(str(machine_number)) [17:09] from juju import placement [17:10] tvansteenburgh: I have another question… is there a command in the sdk for executing the "juju add-machine ssh:" , i was not able to find it [17:11] tvansteenburgh: unfortunately the placement did not fix it https://p.rrbone.net/paste/So9PLneL#CsMGASLtg54WhuTbZkR39baUi-Is5bC4aSh8uPTnAyT [17:12] lorenzotomasini: re: add-machines https://github.com/juju/python-libjuju/issues/51 [17:12] lorenzotomasini: need to see your code [17:13] tvansteenburgh: https://p.rrbone.net/paste/IVbSnOTe#d7jeohncqjo5SN437DmHLP9A5amB4fKe2VBcoi6udlS [17:14] ahhh [17:16] to=[dict(scope='#', directive=str(machine_number))] [17:16] lorenzotomasini: try that ^ [17:21] tvansteenburgh: ok now i get a different error but that is my fault [17:21] tvansteenburgh: thanks [17:29] tvansteenburgh: but…. when i fixed the last error, i got… this one [17:29] https://p.rrbone.net/paste/gccTi3Fr#O7+-2oKOX6V2b6L8nzMRsd+6Agbs1nfc1C5FziEimIM [17:30] hmm === scuttle|afk is now known as scuttlemonkey [17:38] lorenzotomasini: seems to be a bug [17:44] lorenzotomasini: i'll work on a fix for this [17:47] tvansteenburgh: ok thanks, will this be included in an official release? if so, any chances of including the #51 github issue fix? [17:47] lorenzotomasini: yes and yes [17:48] tvansteenburgh: and last and most critical question: when do you think will this be available? [17:48] lorenzotomasini: early next week probably [17:48] :) [17:49] ok thanks [17:49] if you want to hack the libjuju source i can tell you how to work around the placement bug in the meantime [17:50] yeah sure [17:51] lorenzotomasini: comment out these lines https://github.com/juju/python-libjuju/blob/master/juju/model.py#L932-L937 [17:51] tvansteenburgh: I see a placement-fix branch… is it already fixed there? [17:52] lorenzotomasini: no that was just a docstring update [17:52] tvansteenburgh: ah ok, if i comment these two lines it will still deploy on the machine i chose right? [17:52] lorenzotomasini: yeah, it should [18:13] tvansteenburgh: what shoul i put then here? https://github.com/juju/python-libjuju/blob/master/juju/model.py#L1003 [18:14] placement=to, [18:14] yeah [18:34] guys does anyone know where to configure the proxy setting for juju 2.0 [18:34] it use to be like this in 1.2.2 'juju set-env' [18:51] erlon: i believe it's 'juju model-config [http|https|no]-proxy=foo' [18:52] kwmonroe: hmm, thanks ill try [18:53] erlon: i also think you can adjust model defaults so that subsequent add-model calls will use your proxy settings: juju model-defaults *-proxy=foo [18:54] and whilst on the subject, you can see which proxy vars are supported with: juju model-config | grep proxy [18:56] kwmonroe: Hey, when you run java-devenv through cwr-ci, are you getting the "NoneType has no attribute series" error from charm proof still? [18:57] negative cory_fu: http://juju.does-it.net:5000/build_bundle_java_devenv/1/report.html [19:02] kwmonroe: ok, thanks === frankban is now known as frankban|afk [19:36] are we having fun yet Juju world? [19:41] juju deploy lunch --to belly [20:54] seriously, this disconnect crap has got to stop, didn't have this issue on windows [21:37] kwmonroe: I think you were right to push back on "sleep 10" as a work around for the "unremovable model" issue. I can reproduce this on the CLI with `juju add-model foo; juju destroy-model -y foo` [21:37] Looking for an existing bug now [21:39] kwmonroe: https://bugs.launchpad.net/juju/+bug/1635052 [21:39] Bug #1635052: Destroying model shortly after creation results in endless loop

=== mup_ is now known as mup [21:58] cory_fu: thx for finding a faster repro. i do loathe "sleep x" as a fix for anything, except insomnia. [22:04] kwmonroe: Can you test it on the latest beta? [22:09] cory_fu: is rc2 out? [22:09] Not that I'm aware of. Did you still see this on beta5? [22:10] Wait, is rc1 out already? My snap didn't update [22:10] heh, rc2 is out [22:10] read your mail cory_fu! [22:10] ha [22:10] I guess I need to switch from --channel=beta to --channel=candidate [22:29] hey cory_fu, looks like this was fixed in rc1. def fails in 2.0.3, but rc1+ looks ok: http://paste.ubuntu.com/24016323/ [22:30] Good to know [22:31] yeah i guess good for you.. now i've got an undestroyable model hanging around. [22:31] heh [22:31] If it doesn't have any units, it's not really doing any harm. [22:31] it harms my ocd cory_fu [22:32] :) [22:49] does anyone here use the go 1.8rc3 package? if so, how do you get /usr/bin/go to point to the right binary? update-alternatives doesn't seem to know about it. [22:58] is it bad that I am using Ansible to boot strap my Juju environment?!?! [23:04] stormmore: you could shake magnets over a hard disk to bootstrap your juju env and i wouldn't be mad. [23:05] (as long as you don't ask for help) [23:20] mbruzek, lazyPower: how can I specify a ssl key/cert for the kube-api-endpoint? It seems I need the fqdn of my kube-api-endpoint to exist in the SANS of my key/cert used on the kube api endpoint (kubernetes-master). The privately signed key/cert from easyrsa give kubernetes gives me an "untrusted authority" error when trying to register workflows via kube-api-endpoint. I'm trying to specify a publicly signed [23:20] wildcard cert there for my domain, but I think I need to add the kub SANS to it as well, via chain or something. Hoping to try and flesh out what my options are here .... [23:30] oh my ... I just found the work around I think, there is an '--ssl-verify' bool arg that can be fed to deis [23:31] http://paste.ubuntu.com/24016692/ [23:33] kwmonroe, really I am having Ansible bootstrap MaaS as well as Juju since Juju requires an existing controller