[00:03] oy, magicaltrout, you're in dc and you didn't tell me! :) [00:15] not any more marcoceppi sadly [00:15] flew home last friday [00:16] out there helping track down human traffickers and gun smugglers on the dark web... as one does [00:25] magicatrout: we both managed to snub marcoceppi. [00:25] Whoops. [00:26] ^ magicaltrout (without typos in your handle, this time) [00:30] yeah mine involved minimal social activity, and lots of crazy shit i didn't understand [00:30] i have since enrolled myself on a deep learning course [00:37] cool, well if you're ever back in DC hit me up [00:38] i shall be marcoceppi [00:38] probably march [00:38] i'll prod you when i find out [00:38] magicaltrout: cool, hopefully I'll also be in town ;) [00:40] marcoceppi: should we be including a reference to datadog in the interfaces? [00:41] marcoceppi: erggg, *in the datadog state interface name [00:41] jeeezee - s/state/stat/ [00:42] e.g. interface-php-fpm-stats [00:43] should we make it a convention to possibly have 'interface-datadog-' for all of the integrations? [00:43] since there will be alot of them [00:44] I feel like it will be confusing if we don't signify datadog in there somehow ... bc the names of the majority of the integrations are alto the names of juju charms [00:51] bdx: other than datadog being a reference implementation, there's really nothing datadog specific [00:58] marcoceppi: nice, that makes sense.... so then these interfaces could be applicable to other use cases hopefully [00:58] bdx: that's the plan [00:59] bdx: I hope to do a sysdig implementation using a few of these [00:59] as a proof of concept [01:00] ahh ... Sysdig looks like they are blasting native container support [01:00] bdx: yeah, I'm interested in it for kubernetes [01:02] I figured as much ... from a high level ... it looks like they rolled a bunch of cli monitoring tools into one place .. cool ... no gui tho? [01:03] sysdig looks really cool ... I'll have to dig into it [01:07] I see now, sysdig doesn't log data really, it more or less gives you a realtime (and for some things historical) view into the system through a centralized ui - very useful [01:10] bdx: yeah, you can then tap that into sysdig cloud, their saas === CyberJacob is now known as zz_CyberJacob [01:30] sysdig cloud is *intense* [05:20] Hi i am facing browser issue with z ubuntu , getting the error like "Segmentation fault (core dumped)" === stub` is now known as stub === Guest64506 is now known as ahasenack === ahasenack is now known as Guest79511 === frankban|afk is now known as frankban === Mmike_1 is now known as Mmike [09:21] Good morinig Juju world === zz_CyberJacob is now known as CyberJacob [10:03] Good morning ! [10:19] kwmonroe,Hi Good morning [10:20] kwmonroe i have some firefox browser issue, we are charming IBM products [10:21] kwmonroe can you help me regarding the browser issue on ubuntu z machine [11:01] j #EliteBNC [11:02] indeed! [11:04] sticky fingers [11:29] magicaltrout: are you still in the EU? [11:29] magicaltrout: How is the car project going :) [11:30] Ankammarao: kwmonroe will be available later today. If there is anything I can help you with let me know [11:31] kjackal Hi [11:32] kjackal, having issue with firefox in ubuntu z terminal [11:32] yeah kjackal I'm sat in the frozen heartlands.... mostly known as the east of england [11:33] kjackal getting error when run firefox in the terminal "Segmentation fault (core dumped)" [11:34] Ankammarao: is this the case with other browsers as well? [11:35] Ankammarao: does not seem to be a juju related issue. Is it? [11:35] kjackal we have tried with chromium browser to install but getting error [11:36] kjackal yeah its related to ubuntu machine i guess [11:38] Ankammarao: I do not know much about the firefox build on Z [11:38] Ankammarao: however I would recomend you asking in #ubuntu-powerpc [11:39] kjackal thank you [11:39] Ankammarao: here is a list of IRC channels we have. You might find a better match https://wiki.ubuntu.com/IRC/ChannelList [11:40] kjackal i will check it thanks [11:41] magicaltrout: Ahhh very nice. The alternative would have been a life-threatening desert (LA) where coconuts would ruin your brand new tesla :) [11:54] this is true kjackal [12:19] stub: will this work? [12:20] stub: http://paste.ubuntu.com/23588297/ [12:21] stub: mainly, can I set access to 'pg_stat_database' and is that how you would interact with the results [12:21] marcoceppi: That looks correct [12:21] \o/ about to test, so we'll see ;) [12:21] marcoceppi: oh... pg_stat_database is a view inside every database, not a database itself. [12:22] marcoceppi: but it should work... it just creates an empty database called pg_stat_database for your client to connect to. [12:23] marcoceppi: And you probably need to pass in the database name to configure_integration (psql.maser.dbname, which in this case will be 'pg_stat_database' because that is what you asked for) [12:25] stub: according to the setup guide: create user datadog with password ''; grant SELECT ON pg_stat_database to datadog [12:25] trying to replicate this using the pgsql interface, but maybe that's not possible? [12:25] marcoceppi: You would do that after connecting to the database you want to use [12:26] stub: well, I want all databases [12:26] stub: the documentation has no parameter for "database" [12:27] marcoceppi: datadog needs to connect to a database and run selects on the pg_stat_database table. In this case, it doesn't matter what dbname you use. [12:27] marcoceppi: So... [12:27] marcoceppi: In configure_db, I'd set_database('datadog') since that is unlikely to conflict with anything else [12:28] marcoceppi: And ... [12:29] marcoceppi: in configure_postgresql_integration pass 'dbname': pgsql.master.dbname [12:30] marcoceppi: For the charm to do the grant, it needs an administrative connection. And if you give it an administrative connection, it doesn't need the grant run. [12:30] marcoceppi: So just relate it to db-admin instead of db and you might be done [12:33] (I should have called the set_database method set_dbname to avoid confusion. The setting on the relation is 'database', which I inherited. But in the PostgreSQL world it is usually referred to as dbname) [12:33] no worries [12:34] trying with db-admin [12:34] thanks! [12:34] np. === freyes__ is now known as freyes [12:39] marcoceppi: if you get that working, try it with a non-administrative connection. The grant should be unnecessary, at least with modern versions of PostgreSQL. I'm able to query it as an unprivileged user here [12:40] (some views are protected, like pg_stat_activity, but pg_stat_database doesn't contain any sensitive information and seems to be public) [12:41] stub: cool, will give it a go === deanman_ is now known as deanman [12:58] marcoceppi: used the layer-mongodb in our chat charm - but the default 2.6.10 instance needs an `rs.initiate()` from mongo shell before being functional ... is there anyway to set that by default? [13:02] slrocketchat: mongodb charm is missing replicaset support, it's the last thing needed [13:03] marcoceppi: thanks! [13:03] slrocketchat: I've got a stub for it, but I need help completing it, happy to collab on it [13:04] slrocketchat: I started mapping that out here https://github.com/marcoceppi/layer-mongodb/blob/master/lib/charms/layer/mongodb.py#L101 [13:05] I just never was able to get replicasets to come online in a stable fashion [13:06] marcoceppi: happy to take a look. all the docker solution i see uses tricky unreliable time delay strategies. But we only need the single instance as a primary (because Rocket.Chat acts as secondary and parses oplog) so our requirements are non-typical. [13:07] slrocketchat: well, Juju as a third party, does a "leadership" election. I'm happy to pilot a lot of the code, I just need someone with better mongo experience than me to help me figure out the pitfalls [13:14] multicontainer docker stuff, still makes me sad, so i'm not surprised its flaky! ;) [13:15] marcoceppi: we can definitely help to put the layer through the grinder .. once we complete the charm it is expected to be widely deployed in varied configuations :) [13:16] slrocketchat: well I love rocket chat, so happy to work with you all :) [13:16] * magicaltrout googles rocket chat.. i clearly live in a cave [13:17] marcoceppi: goes both ways - we enjoy having gurus around us, solving problems that we hit in seconds with no run arounds :+1: === scuttle|afk is now known as scuttlemonkey [14:39] Hi, quick question, might have stumbled on this before but don't recall. Shouldn't newly created models be inheriting same configuration of default? [14:44] deanman: only if you juju set-model-defaults [14:44] deanman: othewrise juju set-model-config only applies to the currently selected model [14:46] lazyPower, during bootstrapping i configured http proxy and that was passed to defalt. When i created a new one it wasn't inherited. Should i manually configure every new model or is there a better way ? [14:46] deanman: i'm fairly certain the bootstrap constraints only apply at time of bootstrap in juju2. marcoceppi can you fact check my statement above? ^ [14:47] No worries, just want to make sure whether there is a quick way to copy defaults from one model to another. [14:47] deanman: i'm pretty certain you want to set-model-defaults then [14:49] `juju set-model-defaults` does not exist on my 2.0.1 [14:49] newer command addition ? [14:51] got it, `model-defaults` [14:53] no way to pass a file that has pre-configured the key-value configurations? [14:59] deanman: you can [15:00] deanman: you should be able to [15:01] deanman: doesn't look like it has support for feeding a file. That looks like a bug/feature-request to me. [15:03] ok thank you, i''ll file two feature requests then, one to be able to carry over configuration passed during bootstrap and a second to be able to load configuration to a model using a file. [15:05] trying to co-locate two charms on the same machine but i'm getting into trouble with mis-matching python libs - would it be a good idea to try to run charms in separate virtual envs? [15:09] SimonKLB: which charms? [15:09] marcoceppi: my own and cs:~chris.macnaughton/influxdb-4 [15:09] SimonKLB: if they're reactive charms, you can turn on the "enable_venv" option during build time so that their libs (or at least one of their libs) is in a venv [15:09] marcoceppi: neat! :) [15:09] SimonKLB: https://github.com/juju-solutions/layer-basic#layer-configuration [15:10] thanks! [15:10] SimonKLB: cheers [15:16] lazyPower, got them filed as bugs but its the same process for features right? [15:20] deanman: correct [15:20] SimonKLB: did you file [15:20] https://bugs.launchpad.net/charms/+source/elasticsearch/+bug/1646904 ? [15:20] Bug #1646904: ElasticSearch fails at peer-relation-joined [15:27] lazyPower: thats not me === scuttlemonkey is now known as scuttle|afk [15:31] SimonKLB: k, i didnt think so but was just checking === frankban is now known as frankban|afk [20:44] hey any mojo person around? I'm building a spec, and the secrets phase seems to have deposited my secrets as expected, but my charm didn't get configured with the values in the secrets file [22:21] cory_fu: https://github.com/juju-solutions/matrix/issues/36 [22:34] petevg: Replied. But testing bundles from the charm store is no longer supported at all, and you're attempting to test the current directory as a bundle. We should probably improve that error message [22:36] cory_fu: I didn't realize that we were officially dropping support. That's fine .. we just need to revise the tests to reflect. [22:43] get with the times petevg !!! [22:43] magicaltrout: I'm so behind. [22:43] me too.... =/ [22:44] * petevg sheds a single tear [22:44] on the plus side, I have one of my guys working on charms now, so i might actually get some stuff with tests, reviewed etc [22:44] stranger things have happened [22:45] although today I am writing a test suite to test our docker containers...... [22:45] * magicaltrout also sheds a tear [22:45] One step forward ... [22:46] Though hooray for testing, regardless :-) [22:46] lol, as long as I don't have to write... tests.... [22:46] he's working on drill though, I want to get some authentication running in it, but once its done, we'll get some tests in and it reviewed then hopefully you guys can utilising it in your big data bundles [22:47] -ing +e [22:47] bcsaller: If the deploy task encounters a critical error, how can it signal that the test should abort? Returning False makes it hang on the other rules [22:51] magicaltrout: I look forward to playing around with it :-) [22:59] liar [23:16] petevg, bcsaller: https://github.com/juju-solutions/matrix/pull/37 [23:18] cory_fu: code looks good to me. [23:18] Thank you :-) [23:19] Merged. [23:19] petevg: Thanks [23:19] np [23:20] cory_fu, bcsaller: I experimented with integrating the "subordinate okay" stuff more organically into the typing, but I wound up just making monstrosities. [23:20] bcsaller: I went with pre-validating the path instead of aborting more cleanly when the deploy failed, but I'm still curious how we would perform a "clean" abort if we needed to. [23:21] cory_fu: I think i have some outstanding code for that. let me see if it still works [23:21] bcsaller: That also came up when I was working on the ubuntui changes. I was going to add a Cancel button but I couldn't figure out how to make it actually abort [23:25] cory_fu, bcsaller: shouldn't it be something like loop.stop(), model.disconnect(), loop.close()? (I guess you may not have access to the model to disconnect it, and the disconnect method may not work after you call loop.stop ...) [23:40] I see now, sysdig doesn't log data really, it more or less gives you a realtime (and for some things historical) view into the system through a centralized ui - very useful [23:41] aawwwweee