[13:56] <jamespage> SpamapS, m_3 (or anyone else): any objections if I promulgate the hbase charm? it been pretty solid for me for at least a month now...
[14:20] <_mup_> Bug #987853 was filed: Juju Zookeeper node agent does not handle reboots <juju:New> < https://launchpad.net/bugs/987853 >
[14:51] <negronjl> jamespage: I'm ok with it.
[14:51] <jamespage> negronjl, marvellous - ta
[15:59] <jcastro> ok folks, we have a Mac client!
[15:59] <jcastro> thanks to imbrandon! Round of applause please!
[16:00] <SpamapS> jamespage: has nobody reviewed it?
[16:04] <hazmat> bac, ping
[16:05] <hazmat> imbrandon, rockin
[16:05] <bac> hi hazmat, thanks for the patch.  i haven't had a chance to test it yet.  i'll do so right after lunch.
[16:06] <bac> hazmat: i assume that's what you wanted.  :)
[16:07] <imbrandon> :)
[16:07] <hazmat> bac, cool thanks, indeed it was.  i wasn't able to reproduce the original issue, but i've seen reports of it before relating to other cli commands, this should hopefully generically resolve them.. IFF the issue is the zk connection and not the ssh tunnel. if its  the tunnel or ssh process management then it won't help.
[16:08]  * imbrandon did not do too much, basicly the equiv of packaing a rpm/deb, but charm-tools is next and i'm FRACKIN exstatic on cloud 9 about omgubuntu.co.uk right now
[16:08] <bac> hazmat: ok, i'll let you know.  the original is easily reproduced so we should be able to see if it helps.
[16:08] <imbrandon> gotta spam this one line for our deploy a min ago me marcoceppi did : This rush generated 25,661 successful hits in 1.0 min and we transferred 1.98 GB of data in and out of your app. The average hit rate of 414/second translates to about 35,796,365 hits/day.
[16:09]  * imbrandon wants to frame those numbers
[16:09] <imbrandon> thats the juju omg wordpress charm on a single m1.small :)
[16:10] <hazmat> nice
[16:14] <imbrandon> me and marcoceppi gonna work on getting it split up now that there is suborainates into a nginx-proxy nginx wp-install and wp-site, i think is the tentive plan now :)
[17:04] <_mup_> Bug #987930 was filed: MAAS provider - Failure to bootstrap when no port is specified for maas-server <juju:Confirmed> <juju (Ubuntu):Confirmed> < https://launchpad.net/bugs/987930 >
[19:27] <hazmat> DevOp Borat’s words: “Is all fun and game until you are need of put it in production.“
[19:42] <fwereade> hazmat, he's an icon to us all ;)
[20:30] <jcastro> SpamapS: anything to promulgate today?
[20:30] <jcastro> Last call for juju contributor release notes!
[20:31] <marcoceppi> jcastro:  I've got some bandwidth to do a review/promulgate
[20:37] <jcastro> it'd be nice to get drupal done and in
[20:37] <jcastro> imbrandon should totally be in there
[20:37] <imbrandon> ?
[20:38] <imbrandon> oh yea i broke the drupal one a little bit ago, if your finaly gonna prom it give me a few!
[20:39] <imbrandon> jcastro: and i think releaseing the omg one with marcoceppi was a good step too :)
[21:13] <robbiew> jcastro: ping
[21:14] <jcastro> robbiew: yo
[21:24] <adam_g> given the new features that allow hooks to act on other relations, what is the correct way to test that a relation actually exists?
[21:26] <jimbaker> adam_g, use relation-ids
[21:26] <adam_g> jimbaker: thats what i was trying to do. that will give me the list of interfaces, correct?
[21:27] <jimbaker> it will give a list of relation ids for a relation name
[21:27] <adam_g> relation-list seems to list names of related service units, which can be arbitrary depending on how its deployed
[21:27] <jimbaker> where the default is $JUJU_RELATION (if defined, as it would be in a relation hook)
[21:28] <adam_g> jimbaker: so, if i need to test the presence of a relation that provides some specific interface, i'd need to use a combination of relation-list + relation-ids ?
[21:28] <adam_g> (first time using any of this stuff, btw)
[21:29] <jimbaker> adam_g, you would need to something about the charm from metadata.yaml for any interface support
[21:29] <jimbaker> it was intentionally not added in core juju
[21:29] <jimbaker> speaking of non core juju, you might find this command useful: https://code.launchpad.net/~jimbaker/juju-jitsu/juju-do/+merge/103366
[21:30] <jimbaker> adam_g, i'm pretty certain one could readily write a command that parses the metadata, then does the test you asked, with this proposed juju-jitsu addition
[21:31] <adam_g> jimbaker: i need this to happen from within hooks, not outside
[21:32] <adam_g> jimbaker: i essentialy need to something like:  if RELATION_TO_FOO_EXISTS: configure_this_way() else: configure_that_way()
[21:33] <adam_g> jimbaker: since i can't count on service foo being named foo, i should be able to test that a relation to something that provides a known interface to foo exists, right?
[21:34] <jimbaker> adam_g, it's available in the topology, but was excluded from the support i just mentioned. maybe we could pilot something in juju-jitsu for that support, to see if it really makes sense?
[21:37]  * adam_g shrugs
[21:37] <jimbaker> adam_g, curiously enough, hazmat and bcsaller are discussing probably a better approach in #juju-dev
[21:38]  * hazmat unwinds a shrug
[21:38] <jimbaker> which is to convey the interface and the charm name in the relation, as i'm reading the discussion
[21:38] <jimbaker> much like ip addr
[21:39] <hazmat> adam_g, re some specific interface.. from the charm's perspective it knows what interface each rel name corresponds to
[21:39] <hazmat> adam_g, so relation-ids on the rel_name should suffice
[21:40] <hazmat> unless you also want to establish that its been joined at least in which case + rel-list
[21:41] <hazmat> hmm. i really want a list of subordinates
[21:42] <adam_g> hazmat: thats what i had originally thought, something like: for r in `relation-list` ; do relation-ids $r ; done
[21:43] <jimbaker> adam_g, that wouldn't work - you are getting members of the relation in relation-list
[21:43] <jimbaker> but you need a relation name for relation-ids
[21:43] <hazmat> adam_g, you've got your nesting inverted
[21:44] <hazmat> adam_g, so what's the use case?
[21:44] <adam_g> hmm. looks like i need to get a debug-hooks session going to grok this for real
[21:45] <hazmat> adam_g, for example if i'm in mediawiki  and i want to check on mysql from some other hook.. i could go rel_ids=`relation-ids mysql `  and then check if any of those have units.. although realistically in this case its only one.. (from the client side)
[21:45] <hazmat> unless its read-slave
[21:47] <_mup_> juju/relation-ids-whitespace-separated r532 committed by jim.baker@canonical.com
[21:47] <_mup_> Support smart formatting for relation-ids
[21:49] <jimbaker> hazmat, fyi, i need to get the above branch in to properly support things with relation-ids - currently relation-ids only output json
[21:49] <jimbaker> i will get it in shortly - just need a test
[21:51] <hazmat> jimbaker, gotcha
[21:52] <adam_g> hazmat: i'm basically trying to find a good common way of managing config files in this swift rewrite. i'd like to be able to use templated config files that change depending on relations to other services (specifically keystone).
[21:52] <adam_g> so that i can just write out a new template in every hook, instead of updating bits of config
[21:53] <hazmat> adam_g, templates in shell or via puppet?
[21:54] <hazmat> or where those in python?
[21:54] <adam_g> hazmat: python
[21:54] <adam_g> hazmat: just messing around right now to see what works best
[21:55] <jimbaker> adam_g, this seems closer to what is done in charm tools, such as charm splicing
[21:56] <adam_g> jimbaker: ?
[21:56] <jimbaker> in terms of ops on charms
[21:58] <adam_g> jimbaker: i dont know what charm splicing is.
[22:00] <jimbaker> adam_g, i'm not certain what the current state is, but conceptually it allows for one to compose a charm out of a set of mixin charms. most of this need goes away with subordinates, fwiw
[22:13] <_mup_> juju/relation-ids-whitespace-separated r533 committed by jim.baker@canonical.com
[22:13] <_mup_> Add verifying test
[22:19] <_mup_> Bug #988065 was filed: Support smart formatting for relation-ids command <juju:In Progress by jimbaker> < https://launchpad.net/bugs/988065 >
[22:20] <jimbaker> hazmat, can you take a look at the above trivial?
[23:01] <SpamapS> jcastro: I'll hit the charm review queue today
[23:51] <hazmat> jimbaker, done thanks
[23:53] <jimbaker> hazmat, so is that a +1? ;)
[23:54] <hazmat> jimbaker, i dunno, does it do what its supposed to? that's unclear
[23:54] <hazmat> ie. should be newline separate as impl, or space separate per the doc string
[23:54] <hazmat> what's the spec say in regards to this
[23:55] <hazmat> and what's the common shell usage for iteration
[23:55] <jimbaker> hazmat, so shell takes whitespace separated. i think convention is one per line in shell too
[23:55] <jimbaker> the other convention in our code is relation-list, which does one per line
[23:56] <jimbaker> if for example, we have $ids=$(relation-ids db), then this will result in say "db:0 db:1"
[23:57] <hazmat> jimbaker, whitespace separate  .. means space separated
[23:57] <hazmat> to me anyways
[23:58] <hazmat> actually thats the common def
[23:58] <jimbaker> hazmat, it's a good question. my natural inclination was to join with a space, but we do have the existing convention in relation-list
[23:58] <jimbaker> but it's all good