[07:53] <jamespage> gnuoy, re bug 1566688
[07:53] <mup> Bug #1566688: Charm installs different l3 package for same openstack release for installs > trusty <neutron-gateway (Juju Charms Collection):Confirmed for gnuoy> <https://launchpad.net/bugs/1566688>
[07:53] <jamespage> thats correct
[07:53] <jamespage> Ubuntu releases after trusty did not have...
[07:56] <jamespage> openswan
[07:56] <jamespage> which was the default vpn option for the vpn-agent until recently
[07:56] <jamespage> strongswan is now supported as well which is in later releases...
[08:00] <gnuoy> jamespage, so it should be http://paste.ubuntu.com/15642553/
[08:00] <gnuoy> ?
[08:00] <jamespage> gnuoy, maybe
[08:00] <jamespage> tbh I took the approach of disabling post trusty as I had not tested with strongswan
[08:00] <jamespage> and we've sinularly failed to get neutron-vpn-agent into main since then as well
[08:01] <gnuoy> ok, I'll go and ponder, thanks
[08:04] <jamespage> gnuoy, is this blocking something?
[08:08] <gnuoy> jamespage, with the new pause/resume code the services that we register against a config have to actually be accurate rather than the scatter gun register loads and don't worry if they don;t exist
[08:09] <gnuoy> so its just fall out from that
[08:09] <gnuoy> I think I just need to update the amulet tests for my pause/resume branch tbh
[08:09] <gnuoy> I have a if-icehouse-or-greater-its-neutron-vpn-agent clause which is obviously now wrong
[08:10] <gnuoy> s/now//
[08:10] <jamespage> gnuoy, yeah - its if_trusty_neutron_vpn
[08:10] <gnuoy> ack
[08:20] <A-Kaser> Hi
[08:49] <jamespage> coreycb, gnuoy: https://review.openstack.org/#/c/302101/
[08:49] <jamespage> follow up for coreycb's keystone ch-resync for 8.1.0 for stable
[09:26] <D4RKS1D3> Morning
[09:27] <D4RKS1D3> Someone has experience launching services in openstack with juju? I receive this error "ERROR instances not found"
[09:30] <webscholar> Hi all
[09:31] <webscholar> I am trying to install juju2 beta3 on ubuntu 14.04 LTS using the ppa:juju/devel but it installs the juju 1.24 version
[09:31] <webscholar> Can any one share the ppa for juju 2 latest version here ?
[09:34] <webscholar> I am following this
[09:34] <webscholar> https://jujucharms.com/docs/devel/getting-started-general
[11:05] <freak_> anyone available ? need help regarding juju quickstart
[11:06] <freak_> getting error:    juju-quickstart: error: invalid version string: 2.0
[11:07] <freak_> i have installed juju 2
[11:07] <freak_> and quickstart version is 1.3.1
[11:28] <rick_h_> freak_: juju-quickstart is deprecated for juju 2.0.
[11:28] <rick_h_> freak_: juju 2.0 does the work that quickstart did
[11:39] <jamespage> gnuoy, https://review.openstack.org/#/c/301761
[11:39] <jamespage> I think that will do the trick, just need test rig time now....
[11:46] <stub> webscholar: Add ppa:juju/devel and ppa:juju/stable, and install the juju2 package
[13:07] <webscholar> is there a way to mention in bootstrap command which node to use as bootstrap ? When I run juju bootstrap <controller-name> my-maas it bootstraps with its own choice
[13:12] <evilnickveitch> webscholar, if you are using MAAS, you can do it with constraints
[13:19] <evilnickveitch> webscholar, see https://jujucharms.com/docs/devel/charms-constraints and:
[13:19] <evilnickveitch> https://jujucharms.com/docs/devel/reference-constraints
[14:26] <jcastro_> tvansteenburgh, https://github.com/juju/docs/issues/692
[14:53] <stub> Does juju-deployer work with juju2? Interested because amulet, and I'm not sure if juju-deployer needs updating or my environment isn't setup the way it wants.
[14:54] <lazyPower> stub tvansteenburgh is working on it.
[14:54] <tvansteenburgh> stub: tip of deployer works on juju2
[14:54] <stub> ta
[14:54] <tvansteenburgh> stub: same for python-jujuclient
[14:55] <stub> I think the juju-deployer issue has saved me tripping over python-jujuclient ;)
[15:49] <jamespage> coreycb, gnuoy, thedac: can one of you do the honours on https://review.openstack.org/#/c/302101
[15:49] <jamespage> +2 review +1 workflow needed
[15:50] <coreycb> jamespage, on it
[15:51] <jamespage> coreycb, ta
[16:09] <tych0> is there a way to set model config globally?
[16:10] <tych0> like if i always want to set image-stream=daily and default-series=xenial for everything?
[16:10] <Spads> when the agent-status message is "run leader-settings-changed hook", is that prescriptive or descriptive?
[16:10] <Spads> like, is that saying what failed, or advising me on how to un-fail it?
[16:12] <marcoceppi> jcastro_: http://paste.ubuntu.com/15654448/
[16:22] <jamespage> gnuoy, urgh
[16:22] <jamespage> so
[16:22] <jamespage> gnuoy, I just hit the nuance of binding a device to dpdk in that it disappears from /sys/class/net/
[16:23] <jamespage> so re-running the same code post binding results in an empty whitelist of pci devices based on resolution by mac address...
[16:25] <gnuoy> jamespage, do the devices appear elsewhere in the /sys tree?
[19:12] <cory_fu> kwmonroe, c0s: How do you feel about my suggestion in the email about all NameNodes sharing a common ssh key?
[19:13] <cory_fu> Does that seem insecure?  I notice that Big Top has an ssh key checked in to the repo, but I assume that will get replaced by a generated one in a deployment, right?
[19:15] <c0s> cory_fu: are you referring to the yesterday's email?
[19:16] <c0s> cory_fu: in Bigtop there are two different places where we use SSH key
[19:16] <c0s> keys
[19:16] <c0s> 1. for ssh fencing (this one is getting generated, I believe)
[19:16] <c0s> 2. for user hdfs ONLY IF a special test-mode is turned on.
[19:17] <c0s> The latter is static, but is only used in a controlled environment for passwordless ssh, because we need it for some low-level HDFS tests (like block recovery, for instance)
[19:17] <c0s> does it make sense?
[19:18] <cory_fu> Yep
[19:18] <c0s> in general, using static key in production is big no-no, of course
[19:18] <cory_fu> So, for fencing, do you think it's insecure for the namenodes to share a (generated) key?
[19:28] <cory_fu> c0s: I ask because it simplifies the key management quite a bit and if the nodes already have passwordless ssh amongst each other, it doesn't seem any less secure for them to use the same ssh key to do it.  Though, it does mean putting the private key in the leader settings, but again, if you access to that, then you already have access to those boxes
[19:28] <c0s> yeah, I agree
[19:29] <c0s> sharing generated key is perhaps ok.
[19:29] <c0s> Ideally, you'd want some sort of PK management system like kerberos ;)
[19:29] <cory_fu> +1
[19:29] <c0s> but it is way too heavy and overly-sensetive
[19:45] <jcastro_> heya evilnickveitch
[19:46] <evilnickveitch> jcastro, hello, having fun?
[19:46] <jcastro_> I don't have perms to do this, but if you want to take every "this page doesn't have navigation for next and back" bugs for the docs and point them all to one, Marco is going to do it
[19:46] <jcastro_> evilnickveitch, it is wonderful to delete large swaths of things we don't need, yes!
[19:47] <evilnickveitch> jcastro, yes, I am looking forward to killing a lot more stuff tomorrow
[19:48] <evilnickveitch> jcastro, to clarify, not EVERY page is going to have next/back?
[19:48] <evilnickveitch> perhaps i need to see the implementation
[19:54] <jcastro_> yeah so like, are we putting something in the footer or ?
[19:54] <jcastro_> marcoceppi, ^
[19:55] <marcoceppi> jcastro_ evilnickveitch each page, that has peers at the same level, will have a forward backward (when applicable). this will be built using the navigation information and embedded at compile time
[19:55] <marcoceppi> evilnickveitch: I'd support a breadcrumb header that we could interprut at build time to say either breadcrumb: true|false and a sane default to drive this functionality
[19:55] <marcoceppi> but it's basically going to be a new mdx plugin
[19:56] <marcoceppi> evilnickveitch: as an example: https://jujucharms.com/docs/devel/developer-getting-started#install-libraries-and-tools
[19:57] <marcoceppi> at the bottom this would have "Designing your charm ->"
[19:57] <marcoceppi> evilnickveitch jcastro_ the following page, Designing your charm, would have "<- Install libraries and tools                   Writing your charm ->"
[19:58] <marcoceppi> althight, that's a terrible example, since those are anchors
[19:58] <marcoceppi> but that's the idea
[19:58] <evilnickveitch> marcoceppi, I see how that works for the developer docs, as a lot of that is a narrative. Not sure it works so well in other parts
[19:59] <evilnickveitch> but yeah, whatever, I will take a look if you do it
[19:59] <marcoceppi> evilnickveitch: so we would have the top of page config "breadcrumb" set to false
[19:59] <marcoceppi> evilnickveitch: and we'd enable it where it made sense
[19:59] <evilnickveitch> ok
[20:03] <evilnickveitch> marcoceppi, actually, thinking about it...
[20:04] <evilnickveitch> marcoceppi, it would probably save some effort if you just used the metadata
[20:04] <marcoceppi> what metadata?
[20:04] <evilnickveitch> to turn on and off
[20:04] <evilnickveitch> marcoceppi, at the top of each doc
[20:04] <evilnickveitch> there is a line or number of lines
[20:04] <evilnickveitch> (mostly to set the title0
[20:05] <evilnickveitch> any key:value there is interpreted as metadata
[20:05] <evilnickveitch> so it doesn't get processed.
[20:05] <evilnickveitch> hang on...
[20:06] <evilnickveitch> marcoceppi, https://pythonhosted.org/Markdown/extensions/meta_data.html
[20:06] <evilnickveitch> at the moment it is used for the page titles, and the todo generator
[20:06] <marcoceppi> evilnickveitch: cool
[20:06] <evilnickveitch> but it would be simple to add Breadcrumb:true or whatever
[20:06] <evilnickveitch> the metadat is already parsed in the build tool
[20:07] <evilnickveitch> i think you get a dict
[20:07] <evilnickveitch> so, yeah
[20:07] <evilnickveitch> try that way first :)
[20:18] <evilnickveitch> marcoceppi, actually, thinking about it even more, you could have Previous and next instead in the metadata, and use the urls as values. It does mean some manual editing to begin with but would be a lot simpler to implement.
[20:19] <evilnickveitch> anyhow, I think I have done enough thinking for one day
[21:06] <tvansteenburgh> dpb1_, fginther: fyi https://code.launchpad.net/~tvansteenburgh/python-jujuclient/juju2-fixes/+merge/291182
[21:55] <dpb1_> tvansteenburgh: looking
[21:56] <dpb1_> tvansteenburgh: you have an error ~2613 it looks lik
[21:56] <dpb1_> like
[21:57] <tvansteenburgh> dpb1_: jeez, htf did that happen
[21:57] <tvansteenburgh> dpb1_: fixing
[21:57] <dpb1_> :)
[21:59] <tvansteenburgh> dpb1_: fixed and pushed
[22:00] <dpb1_> re-merging
[22:06] <dpb1_> looking now
[23:16] <c0s> kwmonroe: if not with jujuresources how else can I can download binaries specified in resources.yaml? Is there any helper code for that?