=== thumper is now known as thumper-afk === thumper-afk is now known as thumper === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [05:22] hello [05:22] i added one admin address in postgresql service configuration [05:23] can't understand why juju doesn't add corresponding line to pg_hba.conf? [05:23] and when i add it manually it by some reason restores pg_hba.conf file to original state [05:23] i use ubuntu 14-04 [05:23] juju-local [05:23] any ideas? [05:26] or maybe i could do something to add my lines to pg_hba.conf or somewhere else? [05:30] ok, found solution [05:30] entered ip address in CIDR form [05:30] juju is so juju === CyberJacob|Away is now known as CyberJacob === vladk|offline is now known as vladk === CyberJacob is now known as CyberJacob|Away === CyberJacob|Away is now known as CyberJacob === vladk is now known as vladk|offline === vladk|offline is now known as vladk [09:06] yep === vladk is now known as vladk|offline === vladk|offline is now known as vladk === urulama is now known as uru-afk === uru-afk is now known as urulama === CyberJacob is now known as CyberJacob|Away === vladk is now known as vladk|offline === vladk|offline is now known as vladk === vladk is now known as vladk|offline === vladk|offline is now known as vladk === psivaa is now known as psivaa-lunch === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha === psivaa-lunch is now known as psivaa [14:24] asanjar, lazyPower, you guys wanna talk bundles? [14:25] sure [14:26] can I whip one up in jujucharms.com? [14:26] I was thinking of just following your readme, then bundle that as a start? [14:27] jcastro: that'll work - mbruzek is working on getting hive2 promoted into trusty for the store [14:28] asanjar, where would hive fit in? [14:31] hive is used as a bigdata warehousing plaform with SQL like interface. It fits in on top of hdfs for data access and distribution while using mapreduce to process the data based on HiveQL commands. [14:32] so in our solution - we will have Hive connected to HDFS and YARN [14:33] I assume you'll add that to the trusty/hadoop readme as well? [14:33] also we have elasticsearch-hadoop.jar included in HIVE jar path to communicate with ES cluster for data indexing === vladk is now known as vladk|offline [14:34] ok, I'll try to make it less than 9 units [14:34] so it's orangeboxable [14:34] jcastro: which part particularly? the notation about Hive being the big data warehousing? [14:35] ok, why don't I get a bundle up and running that is the scale out hadoop example with ES [14:35] and then you guys tell me what to connect next? === vladk|offline is now known as vladk [14:36] lazyPower: I found this last night http://www.elasticsearch.org/guide/en/elasticsearch/hadoop/current/hive.html [14:37] there are HIVE sample command for ES [14:38] interesting [14:38] would you try them, that would be a good demo for jcastro [14:38] something as simple as this: [14:38] CREATE EXTERNAL TABLE artists (...) [14:38] STORED BY 'org.elasticsearch.hadoop.hive.EsStorageHandler' [14:38] TBLPROPERTIES('es.resource' = 'radio/artists', [14:38] 'es.index.auto.create' = 'false') ; [14:38] would show hive is talking to ES [14:39] jcastro: would that be sufficient [14:39] Sure, give me a bit to wrap what i'm working on [14:39] jcastro: ship me your bundle when you get it mapped up and i'll work through this ES tutorial [14:39] in validation mode [14:39] I'll push it in about 10 minutes [14:40] if jcastro could just do this aspart of the demo: [14:41] CREATE EXTERNAL TABLE artists ( [14:41] id BIGINT, [14:41] name STRING, [14:41] links STRUCT) [14:41] STORED BY 'org.elasticsearch.hadoop.hive.EsStorageHandler' [14:41] TBLPROPERTIES('es.resource' = 'radio/artists'); [14:41] -- insert data to Elasticsearch from another table called 'source' [14:41] INSERT OVERWRITE TABLE artists [14:41] SELECT NULL, s.name, named_struct('url', s.url, 'picture', s.picture) [14:41] FROM source s; [14:42] asanjar, to add elasticsearch [14:42] the readme doesn't mention how to add it to the scale out configuration === vladk is now known as vladk|offline [14:44] http://i.imgur.com/xqjY5r0.png [14:44] this is what I have so far [14:50] asanjar, ^^ what should I connect elasticsearch to? [14:51] jcastro: looks like we have a path for hive. export your bundle and ship it to me so i can deploy it [14:51] i'll start hacking on this ES demo [14:52] ok on it [14:59] lazyPower, https://code.launchpad.net/~jorge/charms/bundles/oscondemo/bundle [15:04] jcastro: got it, give me a few and i'll ping back with results. [15:19] lazyPower: I am in a meeting, will join u as soon as I am done [15:20] asanjar: no rush, i'm just getting started [15:28] weird [15:28] jcastro: drag/drop of the bundle yielded deployment errors - only ES came through [15:28] I only just pushed it [15:28] I haven't tested it [15:28] ack - just a heads up [15:28] Got pulled into a call [15:28] i'll get it working and push a patch once its deploying proper [15:29] what errors did you get? [15:29] the gui didnt say, just errors on bundle [15:29] I was just following the readme [15:29] oh, use deployer man. :) [15:30] :-) [15:30] :P [15:30] * lazyPower pushes the easy button [15:31] same status result [15:31] pastebin please [15:31] An error occurred while deploying the bundle: no further details can be provided about a minute ago [15:32] thats all i get. i'm going to nuke the relations and see if its the service definition [15:33] oh I also forgot to rename envExport to something useful [15:34] yeah i took care of that === vladk|offline is now known as vladk [17:01] lazyPower: go back to "eco" === urulama is now known as uru-afk [17:12] wow [17:12] lazyPower, my deployer straight up crashes on that bundle [17:13] jcastro: i'm rebuilding it by hand [17:13] this deployment configuration specified is also not one you'll see very commonly [17:14] we're going to hand off a ready made deployment W/ elasticsearch and hive [17:14] i'll post a bzr branch shortly for you to consume [17:16] ack, I'm going to lunch === ev_ is now known as ev [18:04] anyone experiencing 409 conflicts with MAAS+JUJU when adding machines? === scuttle` is now known as scuttle|afk [18:14] l1fe: that usually means no MAAS nodes match your constraints (or you're out of available nodes) [18:18] right, in this case, all nodes are ready [18:18] (well aside from the bootstrapped node) [18:18] l1fe: what architecture are they? [18:18] amd64 and trusty [18:18] the environment actually used to work [18:18] and then after one too many destroy-environments [18:18] l1fe: run with --debug see what juju is saying maas is complaining about [18:19] jcastro: http://i.imgur.com/DZfKZWb.png [18:19] 409 is pretty much the MAAS API saying "I heard you, but I don't really feel like doing this because: ..." [18:19] i can ship you the bundle as is - we're still validating [18:19] it's pretty generic [18:19] 2014-07-11 16:58:06 ERROR juju.provisioner provisioner_task.go:421 cannot start instance for machine "1": cannot run instances: gomaasapi: got error back from server: 409 CONFLICT (No matching node is available.) === roadmr is now known as roadmr_afk [18:20] i've even reprovisioned the entire cluster [18:20] l1fe: yeah, so maas says you don't have any nods available. What size ram are these? [18:20] re-installed maas and juju [18:20] all 16gb [18:20] l1fe: are you on 1.20 ? [18:20] 1.20.1 [18:20] * marcoceppi scratches head [18:20] and it literally was working 24 hours ago [18:21] until after the last fateful destroy environment [18:21] it's always that last destroy ;) [18:21] :) [18:21] And you've tried recommissioniong the nodes? [18:21] yup [18:21] full recommission [18:21] reason why i even destroyed my environment [18:21] it's weird [18:22] was after i reported https://bugs.launchpad.net/bugs/1340261 [18:22] <_mup_> Bug #1340261: juju add-machine lxc:0 fails to start due to incorrect network name in trusty config

[18:22] that it bootstraps [18:22] jcastro: lp:~lazypower/charms/bundles/oscondemo/bundle/ [18:22] i wanted to re-setup my openstack environment with lxc...and figured i'd just start from scratch [18:22] yeup, bootstraps...and if i bootstrap, it can even pick up random nodes [18:22] that it would otherwise NOT pick up during add-machine [18:23] l1fe: what does juju get-contstraints say? [18:23] oh jesus [18:23] you're kidding me [18:23] as;dlkfjas;dlfkjlasdf [18:23] thanks marcoceppi :) [18:24] <3 [18:24] l1fe: was it set to 32 bit? [18:24] when i was bootstrapping, i was setting a constraint for 32G because i wanted to install on a particular maas node [18:24] ah, yeah [18:24] that sets it globally [18:24] that's good to know [18:24] omg [18:24] i did a full reprovision twice [18:24] thanks [18:25] l1fe: no problem, you can use juju set-contstraints to put the mem limit lower [18:25] yup [18:26] i just did juju set constraints "" [18:26] to clear it [18:26] cool [18:26] oh my god, i thought i was going insane [18:26] i went down to 1.18 [18:26] and 1.19 in devel [18:26] and was like...this was working!! [18:39] lazyPower: https://bugs.launchpad.net/charms/+source/mysql/+bug/965094 [18:39] <_mup_> Bug #965094: Provide config option to specify database encoding [18:40] jrwren: thanks for filing the bug. The charm maintainer will see it shortly. [18:42] lazyPower: i want to fix it. i wanted to ask you about possible ways to fix. [19:15] jrwren: the "problem" with setting it in the interface, is that redefines the interface [19:15] jrwren: it' a good idea, but the new key will have to be 100% optional [19:15] jrwren: so, you'd want to have a default-encoding configuration option, which is used when no encoding is sent via the relation [19:16] agreed. going for 100% optional [19:16] exactly what I was thinking. [19:17] i spiked it and tested it out and it works, i'm just charm new, and wanted to ask before i submit a merge request [19:26] jrwren: :) we're really open to code reviews as MP's [19:26] it makes life great when its of good quality and we just merge it in [19:26] if not, its a source to track the conversation as the merge evolves [19:27] lazyPower: ok, thanks. i'll get something up [19:32] quick question regarding juju and lxc containers...does juju provide a unified network fabric for any services exposed over LXC? as in something on LXC:0 being able to communicate with LXC:1 [19:33] l1fe: yes, something in LXC:0 will be able to talk to LXC:1 [19:33] hmm [19:33] so, i have a mysql lxc on 1/lxc/1 [19:34] and nova-cloud-controller on 0/lxc/1 [19:34] and i get errors where it can't connect to mysql [19:34] via the ip address [19:34] which makes sense since those 10.0.* ip addresses are only for each machine [19:35] physical machine [19:35] unless i'm misunderstanding something [19:36] (given the constraints thing earlier, that's entirely possible) [19:37] l1fe: are you sure teh mysql service has started? its using a bridged virtual ethernet adapter to do the communication in LXC [19:37] LXCBR0 is the default virtual device for lxc connections, and all lxc containers use 10.0.3.x ip addresses by default as well [19:37] yeah, it's started [19:39] yeah, i raised a bug about how lxcbr0 wasn't actually being used, so i even manually went in and changed that in containers directory [19:39] for lxc.cfg [19:41] strange, works on every other setup. The only time i had to change the br interface was when i wanted to bridge it with an actual ethernet device [19:41] and even then i think i configured it wrong and i had soem races where the interface wouldn't come online during boot and I had to manually initialize it every time the server was reset. === vladk is now known as vladk|offline [19:41] but thats a story for another time.... === vladk|offline is now known as vladk [19:42] so if you can reach the MySQL node, and you've verified the server is up - whats the log output that's stating you cannot connec to MySQL? [19:42] https://bugs.launchpad.net/bugs/1340261 [19:42] <_mup_> Bug #1340261: juju add-machine lxc:0 fails to start due to incorrect network name in trusty config

[19:42] for the templates not being setup properly for trusty [19:43] ahhh ok so this is not a local deployment [19:43] this is lxc, in a maas node for density based deployments [19:43] ack. [19:44] unit-nova-cloud-controller-0: 2014-07-11 19:12:44 ERROR juju.worker.uniter uniter.go:486 hook failed: exit status 1 unit-cinder-0: 2014-07-11 19:12:55 INFO shared-db-relation-changed 2014-07-11 19:12:55.604 18833 CRITICAL cinder [-] OperationalError: (OperationalError) (2003, "Can't connect to MySQL server on '10.0.3.34' (113)") None None [19:44] yeah, not local [19:44] sorry forgot to mention that [19:44] these are different physical maas nodes [19:44] its ok, its my own brain going immediately to local when someone mentions LXC [19:44] i'm just trying to test out the best ways to deploy my environment :) [19:45] and experimenting with encapsulating services [19:46] i guess bridging networks between lxc containers across different physical nodes would be something that's not possible right now [19:47] i know that the networking when using juju deploy --to lxc:# can be a bit troublesome when it comes to networking configuration [19:47] ah [19:47] i don't see any DHCP or DNS changes in MAAS (which is controlling all of that stuff right now) === roadmr_afk is now known as roadmr [19:48] There was a topic on this not long ago [19:48] and juju is using the actual IP address vs a fqdn [19:48] there's either a script or a subordinate charm that helps this... i forget how long ago the thread was - but its in the juju mailng list. [19:48] *mailing list [19:49] ah, i'll have to do some searching === vladk is now known as vladk|offline === cmagina_ is now known as cmagina === scuttle|afk is now known as scuttlemonkey === CyberJacob|Away is now known as CyberJacob === CyberJacob is now known as CyberJacob|Away