[00:20] <bdx0> Hey what's up everyone? I'm trying to configure ha for some services by putting them behind haproxy
[00:21] <bdx0> First I set a vip for the service, then make the relation to the hacluster charm
[00:22] <bdx0> After the status settles and becomes idol I add more service units
[00:24] <bdx0> This seems to be a pretty straight forward process .....but I am bleeding caveats, bugs and errors around every corner.....is clustering services behind haproxy known to be problematic to some extent?
[00:33] <bdx> core, dev, charmers, openstack-charmers: charm store is down!
[00:33] <blahdeblah> bdx: Known issue - working on it
[00:34] <bdx> http://paste.ubuntu.com/12871528/
[08:35] <gnuoy`> jamespage, could you take a look at https://code.launchpad.net/~gnuoy/charm-helpers/max_map_count/+merge/274979 if you get a mo?
[08:35] <jamespage> gnuoy`, +1
[08:35] <gnuoy`> thanks
[10:41] <admcleod-> is there a way to flush the action queue for an agent?
[10:44] <admcleod->  i moved /var/lib/juju/agents/unit-plugin-0/state/uniter but i suspect thats not the best way since now its runningthe install hook again
[12:03] <Icey> any idea why https://code.launchpad.net/~chris.macnaughton/charms/trusty/telegraf/trunk (last pushed to LP 17 hours ago) isn't yet available in the charm store at cs:~chris.macnaughton/trusty/telegraf ?
[12:15] <rick_h_> Icey: does charm proof pass on it?
[12:16] <rick_h_> Icey: w/o error?
[12:17] <Icey> eh, 3 W, 2 I; would be nice if passing charm proof was listed as a requirement on the actual documentation for submitting a charm to the charm store in a user namespace :)
[12:17] <Icey> will get it passing
[12:18] <rick_h_> Icey: just errors have to be clear. You're correct that it should be documented on that. It's step 4 on https://jujucharms.com/docs/stable/authors-charm-store
[12:19] <Icey> which is not in the section on "Name Space Charms" (https://jujucharms.com/docs/stable/authors-charm-store#name-space-charms)
[12:19] <Icey> :(
[12:19] <Icey> oh well :)
[12:19] <rick_h_> Icey: ok, will file a bug to get that updated. Thanks for pointing me where you were looking
[12:20] <Icey> blech: W: Maintainer format should be "Name <Email>", not "Chris MacNaughton <chris.macnaughton@canonical.com>"
[12:20] <Icey> parsing fail?
[12:21] <rick_h_> Icey: https://github.com/juju/docs/issues/708 filed if you'd like to track
[12:21] <Icey> heh cool
[12:21] <Icey> thanks :)
[12:21] <rick_h_> Icey: hmm, no errors though
[12:25] <Icey> apparently charms with warning aren't passed on to the charm store
[12:25] <Icey> which is kind of annoying because apparently it really doesn't like my email?
[12:26] <rick_h_> Icey: looking, I think this might be due to the legacy (older store) running an older proof. I'm installing the older version now to check if it throws errors
[12:26] <rick_h_> Icey: I know the store only skips charms for error conditions
[12:26] <rick_h_> Icey: if it's not that, then some other bug and we'll get the team to look into it
[12:26] <Icey> https://jujucharms.com/docs/devel/tools-charm-tools
[12:26] <Icey> warning level blocks charm store upload as well
[12:27] <Icey> "Any charm with a Warning or Error will not pass charm store review policy."
[12:27] <rick_h_> Icey: that's for promulgation
[12:27] <rick_h_> Icey: for being recommended, not personal namespace
[12:27] <Icey> that's in the proof docs
[12:27] <Icey> -_-
[12:27] <Icey> I'm cleaning up the proof issues anyways
[12:27] <rick_h_> if you want to be the official telegraf charm in the store and able to juju deploy telegraf it can't have warnings
[12:27] <Icey> but annoying that it doesn't like my name + email format -_-
[12:28] <Icey> that's the goal soon, but not just yet ;-)
[12:28] <rick_h_> Icey: one step at a time :)
[12:28] <Icey> yeah, first official charm is the goal with this, and it's an auxiliary charm for something we're working on :-P
[12:31] <rick_h_> Icey: filed https://github.com/CanonicalLtd/jujucharms.com/issues/168 and ping'd the team to peek at it. We'll see what they find
[12:31] <Icey> apparently tags is bad?
[12:31] <Icey> well, we'll see :-P
[12:32] <rick_h_> no, I know other charms have tags
[12:32] <rick_h_> https://api.jujucharms.com/charmstore/v4/trusty/rabbitmq-server-38/archive/metadata.yaml for instance
[12:33] <rick_h_> so I think I'm pulling at straws
[12:33] <Icey> different format?
[12:33] <Icey> I'm using the yaml array format, rabbitmq is using [] ?
[12:33] <rick_h_> no, i tried to match taht format locally and it still errored
[12:33] <Icey> gah
[12:33] <Icey> well, I'm going to push a change that clears most of the rest
[12:33] <Icey> maintainer had a typo, missing the closing > -_-
[12:33] <rick_h_> Icey: cool thanks, the folks will investigate
[12:34] <Icey> and who knows, maybe it'll go now with less charm proof complaints ;-)
[12:34] <rick_h_> Icey: there's a chance
[13:04] <Icey> rick_h_ I bet if you tried `charm proof` on the version that threw that error before that it'd pass now :)
[13:28] <Odd_Bloke> A review/merge of https://code.launchpad.net/~daniel-thewatkins/charms/trusty/ubuntu-repository-cache/nrpe-servicegroups/+merge/275009 would be much appreciated.
[13:53] <mbruzek> ping rick_h_
[14:06] <rick_h_> mbruzek: pong
[14:07] <mbruzek> I have found my answer in a bug you wrote against docs.  But who (besides you) knows about the ingest process/rules ?
[14:11] <rick_h_> mbruzek: urulama and bac in particular
[14:12] <mbruzek> rick_h_: Thank you.
[15:29] <Odd_Bloke> marcoceppi: Thanks for the merge!
[17:59] <bdx> Whats up charmers, openstack-charmers, core, dev?
[18:01] <bdx> core, dev, charmers, openstack-charmers: I am expeimenting with deploying openstack stateless services behind hacluster, and am experiencing a plethora of issues....do you guys know that ha deploys are entirely borked???
[18:05] <bdx> core, dev, charmers, openstack-charmers: I can exploit and reproduce each problem, and create bugs for each, unless this is being delt with on a higher level somehow/someway.....
[18:06] <lp|lunch> bdx, o/
[18:07] <lp|lunch> bdx, thats interesting, I'm fairly certain we test the OpenStack bundles with an HA deployment
[18:07] <bdx> lazyPower: how sure?
[18:08] <lazyPower> lets take the pepsi challenge
[18:08] <lazyPower> beisner, o/
[18:08] <lazyPower> beisner, we test openstack with hacluster + test HA deployments in osci don't we?
[18:10] <beisner> o/ lazyPower
[18:10] <beisner> lazyPower, yep, uosci exercises HA deploys, but even better ... the whole test infra runs on a juju-deployed trusty-kilo HA openstack cloud ;-)
[18:11] <bdx> lazyPower, beisner: even though my juju env looks great: http://paste.ubuntu.com/12878545/
[18:11] <bdx> the functionality of the openstack services is so broken
[18:12] <bdx> using percona, things are 10x more broken
[18:13] <bdx> hooks error for shared-db for glance, cinder, nova-cloud-controller, and neutron-api if percona-cluster is used
[18:15] <lazyPower> bdx, filing bugs for these issues that you're uncovering would be helpful. This may or may not be a gap in our testing coverage.
[18:15] <lazyPower> but having a bug to track them as they get investigated would be helpful so we have a reference to begin looking at the issue(s) afoot here and we can get this prioritized
[18:16] <bdx> lazyPower: entirely....I will set aside my upcoming weekend to do so
[18:17] <beisner> bdx, there are perhaps infinite topology and ha decisions to make.  here is what we exercise, with percona cluster, essentially 3 units of every service...
[18:18] <beisner> have been running something very similar to that in prod for 1.5+ yrs, and have taken that prod cloud from trusty-icehouse to trusty-juno to trusty-kilo
[18:18] <beisner> these links are dev/testing, but may provide more insight
[18:18] <beisner> but you'll have to self-parse branches into the bundle
[18:19] <beisner> "trusty-kilo-ha":  http://bazaar.launchpad.net/~ost-maintainers/openstack-mojo-specs/mojo-openstack-specs/view/head:/helper/bundles/ha.yaml
[18:19] <beisner> +
[18:19] <beisner> branches:  http://bazaar.launchpad.net/~ost-maintainers/openstack-mojo-specs/mojo-openstack-specs/view/head:/helper/collect/collect-next-ha
[18:19] <beisner> ie.  juju-deployer -B -v -c ha.yaml -d trusty-kilo-ha
[18:20] <beisner> our automation parses the two and puts branches where they belong
[18:22] <beisner> bdx, bear in mind, that the 15.10 openstack charm release is upon us this week.  that means all of these "next" branches will become charm store charm versions very shortly.
[18:23] <bdx> beisner: what I'm getting from this is that alot of the problems I'm having have most likely been delt with in the 15.10 release
[18:24] <beisner> bdx, some bug fixes, sure.  but i'm sitting on a cloud that has existed for > 1.5 yrs, originally deployed with old charms, and upgraded along the way.
[18:25] <beisner> bdx, the big gain with 15.07 and 15.10 are improvements around service status (better feedback in juju stat) + leadership election (juju 1.24.x really helps facilitate clustering activities).
[18:34] <bdx> beisner: sweet. thanks!
[18:36] <beisner> bdx, youre sure welcome!   i'd say have a look at the topology, configurations and usage in that example link and ignore the fact that it references dev charms.
[18:38] <bdx> beisner: entirely....
[18:38] <beisner> bdx, lazyPower - perhaps one day we will have a proper example bundle in the charmstore for openstack HA.  i do think that is on our list-o-stuff to do.
[18:38] <beisner> it's difficult though, since there are so many variables with HA, and one has to pre-determine VIP addresses, etc etc
[18:38] <bdx> awesome....totally
[18:43] <beisner> bdx, oh so one thing to beware of - that example will use a zillion machines.  you would have to co-locate in a more sensible hardware allocation.  such as:  http://paste.ubuntu.com/12878773/
[19:30] <Icey> rick_h_ https://code.launchpad.net/~chris.macnaughton/charms/trusty/telegraf/trunk is still not in jujucharms.com -_-
[19:55] <lazyPower> Icey, does it pass proof?
[19:55] <lazyPower> nothing above an I: will get ingested
[19:55] <lazyPower> as the return code is 1, so it bails out
[19:56] <Icey> % charm proof
[19:56] <Icey> I: File config.yaml not found.
[19:56] <Icey> I is the only issue
[19:56] <lazyPower> can you open an issue here? https://github.com/CanonicalLtd/jujucharms.com/issues
[19:57] <lazyPower> link @ your lp branch, and ping in #juju-gui. this went through rick_h_ before but i'm pretty sure this now is under the maintenance of urulama
[19:57] <Icey> https://github.com/CanonicalLtd/jujucharms.com/issues/168
[19:57] <lazyPower> oh
[19:57] <lazyPower> well then
[19:57] <Icey> :)
[19:57] <lazyPower> i just pinged two ppl, lets hope they notice :D
[19:58] <lazyPower> interesting on that bug though
[19:58] <urulama> hey
[19:58] <Icey> yeah, I made some changes that clean up most of the other charm proof issues
[19:58] <lazyPower> mbruzek, did you see this?
[19:58]  * urulama reads back
[19:58] <lazyPower> mbruzek, issue #168
[19:58] <lazyPower> mbruzek, sorry link is here: https://github.com/CanonicalLtd/jujucharms.com/issues/168
[19:58] <mbruzek> looking
[20:01] <mbruzek> lazyPower: I talked with rick_h_ today about this.  Charms will still get ingested if they have proof warnings, but bundles will fail ingestion with proof problems.
[20:02] <Icey> there are no proof problems though...
[20:23] <urulama> Icey: we'll look into it. the charm seems ok. I'll dig into the logs.
[20:24] <Icey> thanks :)
[20:36] <urulama> Icey: your charm is ok and will get published in charm store
[20:36] <Icey> thanks :)
[20:36] <Icey> what was the issue?
[20:36] <urulama> Icey: there are some technical difficulties with it at the moment, thanks for your patience
[20:36] <urulama> Icey: no charm got published today