[01:20] anastasiamac, axw, menn0: is someone(tm) going to take care of the juju-mongodb SRU verification? https://bugs.launchpad.net/ubuntu/+source/juju-mongodb3.2/+bug/1673652 [01:20] Bug #1673652: SRU juju-mongodb 3.2.12-0ubuntu1 to yakkety and xenial [01:22] mwhudson: veebers is looking at this today \o/ [01:22] anastasiamac: awesome [01:46] anastasiamac, mwhudson we started looking at this today, not sure how long it will take but we should get some testing done in the next day or so [01:46] veebers: i'm not in a hurry at all, i just wanted to make sure it wasn't forgotten about [01:46] veebers: this is great! thank you :D [01:47] mwhudson: we would really really like it, so very unlikely to forget :) m thrilled that u r following up \o/ [04:01] what are the "payment forms" that are being added to juju-gui? [04:02] in this MR https://github.com/juju/juju-gui/pull/2675 [08:07] Hello [08:07] I want to write amulet test case for my charm [08:07] hello [13:21] i want to write amulet test case for my charm [13:22] is there any way to pass resource to this charm From amulet framework [08:07] is there any way to pass resource to this charm From amulet framework [09:40] morning [09:59] Lalit: subprocess.check_call(['juju', 'attach', ....) if you can attach the resource after the initial deployment [10:00] Lalit: If you need the resource attached during deployment, I don't know. That would actually be a juju-deployer thing, which amulet is using for the initial deployment. [15:07] Anyopne here from the community team ? [15:09] CoderEurope: we are all part of the community team, :-) anything I can help with? [15:12] https://api.jujucharms.com/identity [15:12] red alert [15:13] http://paste.ubuntu.com/24275054/ [15:13] I can't seem to login, register, or use juju from any of my clients [15:14] are others seeing this too? [15:14] rick_h:^ [15:15] anyone:^ [15:15] bdx: checking, was just doing stuff before this call started. [15:16] bdx: yes, I see it. Just getting a response going [15:16] none of my juju clients are working ... http://imgur.com/a/betbJ [15:16] ok [15:28] bdx: yes, seems the identity service is not responding. Folks are on top of it [15:28] rick_h: thx [15:30] arosales: sorted now thankyou :D [15:30] CoderEurope: glad its sorted :-) [15:53] random q: did juju2 deliver anything in terms of the 'virtual' service concept? i.e. deploying a shim charm on the controller, or similar? [15:54] bloodearnest: what's a "shim charm" ? [15:54] * rick_h is working to catch up [16:06] bloodearnest: rick_h, i think that's the concept of a "proxy charm", whereby you'd deploy a lightweight doohickey that proxies requests for an existing application that is not charmed. the idea being you could have an existing mysql db that you want juju charms to interact with even though your mysql server is not actually a part of the deployment. [16:07] i know there was work done around that last cycle, but do not know any current status for it [16:07] kwmonroe: yea, nothing new in juju2 towards that. [16:07] sounds like rick_h just got weekend plans ;) [16:07] lol [16:07] my weekend plans involve driving across the country with a trailer :P [16:08] oof! good luck out there! watch out for self-driving ubers [16:08] bloodearnest: just build it into the charm, let me see if I can dig up an example for you [16:09] kwmonroe, that's what I'm talking about, proxy charm [16:09] bloodearnest: https://gist.github.com/jamesbeedy/dc0add471b086e918d2b64c7768d706c [16:11] bloodearnest: like bdx mentioned, some charms handle external entities already (https://jujucharms.com/rsyslog/ comes to mind, see the "Forwarding logs to a system outside of the Juju environment" section). [16:11] so, basically I do a manual db check, and set the state 'prm.manual.database.available' if the user has specified the db -> https://gist.github.com/jamesbeedy/dc0add471b086e918d2b64c7768d706c#file-prm_base-py-L45 [16:11] kwmonroe, bdx: right, that's what we've done forever, and it has drawbacks [16:12] bloodearnest: but I also allow for a database via juju relation [16:12] def set_block [16:12] https://gist.github.com/jamesbeedy/dc0add471b086e918d2b64c7768d706c#file-prm_base-py-L99 [16:12] bdx: that's slick -- using config to handle external db bits. [16:12] kwmonroe: yeah, then the when_any https://gist.github.com/jamesbeedy/dc0add471b086e918d2b64c7768d706c#file-prm_base-py-L129 [16:13] I basically want a single reusable charm that acts as an in-model, relatable, point of access to external services (dbs, logging, metrics, etc) [16:13] so I can remove the relation https://gist.github.com/jamesbeedy/dc0add471b086e918d2b64c7768d706c#file-prm_base-py-L169, and modify the config https://gist.github.com/jamesbeedy/dc0add471b086e918d2b64c7768d706c#file-prm_base-py-L164 [16:13] very cool bdx [16:13] I have such a charm, but I'm forced to deploy it on a wasted vm [16:13] kwmonroe, bloodearnest: then block when none are available https://gist.github.com/jamesbeedy/dc0add471b086e918d2b64c7768d706c#file-prm_base-py-L50 [16:14] my next additions will include disallowing both to be configured at the same time [16:14] kwmonroe: thx! [16:14] bloodearnest: "juju deploy --to " would save you the wasted vm, assuming you have an existing machine that you don't mind jamming your proxy charm onto. [17:38] hey arosales you there so you can stop spamming me? ;) [17:39] magicaltrout: hello :-) [17:39] magicaltrout: well if your license would work with your charms I wouldn't have to spam you [17:39] sorry arosales I'm at a DARPA workshop launch stuff so I'm a bit sporadic in my sitting at a desk [17:39] if you have a license [17:40] you should be able to visit http://yourserver/upload.htl [17:40] magicaltrout: sorry for the bother I figured you were on a project or I hoped maybe even taking a vacation [17:40] html [17:40] you should get an upload form for a license [17:40] magicaltrout: yes tried that, but I can't login with admin/admin [17:40] * arosales posted at https://groups.google.com/a/saiku.meteorite.bi/d/msg/user/6mDNTdwwJEg/EfkWhzgPDAAJ [17:41] I am tearing down my controller and starting fresh just to be sure [17:41] I'll create a new license with this new model since the IP will change [17:41] fair enough, give me access and i'll take a look when you get it stood back up [17:42] coming up as we speak should be about another 10 min [17:42] cool [17:43] i'm around all afternoon, I left the event, DARPA can do defence research but not WIFI it appears ;) [17:45] security risk -- all smoke signals now [17:50] magicaltrout: no luck [17:50] magicaltrout: http://34.223.249.229/upload.html [17:50] selecte upload license; select new license; then stuck in login in loop [17:54] 6 minutes to the Juju Show #9 [17:54] rick_h: you have a link handy for folks who want to join live? [17:55] arosales: https://hangouts.google.com/hangouts/_/mqwvxdrugbaspk7bc7twmrx5wye to join the hangout and I'll get the viewing link [17:55] rick_h: thanks [17:55] https://www.youtube.com/watch?v=FQv-ScdYQeg to watch the stream live [17:55] come one, come all...it's another week of fun [17:59] rick_h 🤣 [18:00] arosales: you coming in? [18:00] arosales: any of the regulars around today or all off at kubecon? [18:03] bdx: please check if all is well for your models in JAAS [18:03] bdx: should be up and running now [18:04] rick_h: see you now :) [18:09] stokachu: ping for ^ [18:09] rick_h: whats the url? [18:10] https://hangouts.google.com/hangouts/_/mqwvxdrugbaspk7bc7twmrx5wye [18:10] urulama: looks like I'm back in business! thanks! [18:10] bdx: sorry for the long outage :( [18:10] QUESTION: Who do we prompt to update jujucharms.com pages ? | Would there be a repo, for jujucharms.com ? | or where best should I go to chat about this ? [18:11] Boat_club: you can file an issue on http://github.com/CanonicalLtd/jujucharms.com/issues [18:11] urulama: cheers :) [18:12] Boat_club: all feedback or suggestions appreciated [18:12] urulama: np! glad its back up now! thx === frankban is now known as frankban|afk [18:17] urulama: Congrats on your 400th issue closed ! [18:18] Boat_club lol, that's only one repo :) [18:19] * Boat_club sips her coffee. [18:20] QUESTION: Are there any issues with juju and systemd ? IDK much about it, but just thought I would ask a silly question :) [18:22] Boat_club: sorry, missed your question. There's no issues that we're aware of [18:23] Boat_club: Yes, it might not play nice if you use partitions: https://bugs.launchpad.net/juju/+bug/1634390 [18:23] Bug #1634390: jujud services not starting after reboot when /var is on separate partition [18:23] doh, that's right [18:23] good call zeestrat [18:23] Hoping it will get fixed in 2.2 :) [18:24] rick_h: okay fantastico show, pal o/ [18:24] Boat_club: thanks for watching [18:24] * rick_h goes to get notes together and onto the video wheee [18:26] rick_h: thanks for charing another juju show [18:26] arosales: np, ty for joining in [18:26] arosales: will have to get some updates from the folks at kubecon maybe next week. curious what's up over there [18:27] ya there is some good kubernetes 1.6 bits coming down the pipe, and I would also be interested to hear how the conf went [18:27] rick_h: guess juju works well on macOSX (?) | macbookair - not so much. [18:29] Boat_club: so it's the client and works well [18:29] Boat_club: I've been using it exclusively for a few days [18:29] Boat_club: I migrated from a 2013 air to this new machine yesterday so definitely works on an air [18:29] rick_h: just need to update this: http://bit.ly/macbookair [18:29] whoopsie [18:30] http://bit.ly/macbookairUbuntu [18:30] Boat_club: hah [18:30] well Juju is across platform. Ubuntu is a platform. :) [18:31] rick_h: cool beans :D [18:45] hey, could anyone from the charmers team pm me? [18:47] kwmonroe: http://interfaces.juju.solutions/layer/nvidia-cuda/ may be interesting for tensor on spark [18:48] looks like tvansteenburgh used this layer in his work on GPUs for k8 @ https://github.com/kubernetes/kubernetes/compare/master...tvansteenburgh:gpu-support [18:49] the other heavy lifting in k8 was the privileged container peices [18:50] someone needs to charm up jupyterhub as well thats looking pretty cool these days [18:52] * arosales thought we had a jupyter charm . . . [18:52] https://jujucharms.com/u/tengu-team/jupyter-notebook-spark/2 [18:53] Jupyter Notebook from tengu [18:53] arosales: Did you notice that www.reddit.com/r/LinuxActionShow was no more ? [18:54] * arosales looks [18:54] * arosales wasn't aware [18:54] arosales: you triggered me at jupiter :D [18:54] ah [18:55] Jupiter Broadcasting [18:57] the hub is a multiuser thing [18:57] magicaltrout: seems https://github.com/jupyterhub/jupyterhub mainly provides multi-user support for the jupyter notebook, correct? [18:57] ok [18:57] as opposed to a stand alone notebook [18:58] may be good to ping seb on that when he is around to see if they have a use case to extend their current notebook charm [18:58] * arosales doesn't see sebastien here atm [19:31] whoa arosales - nice find re: cuda layer. who knew tvansteenburgh was a secret bash coder?!?! [19:33] you never know what lurks under those shades [19:35] I think he was building off SaMnCo work, but ya I am surprised there wasn't a re-factor to python :-) [19:35] that cuda-layer looks awesome - I will be putting it to use soon actually, as one of our applications is now transcoding video and needs hw accel [19:37] when you're stuck in an AirBnB is 3:40 too early for beer? [19:38] sorry I don't understand the question [19:38] too early for beer? [19:38] hehe [19:38] i have an "infrastructure transition discussion" in 20 mins [19:39] i might need a bottle of moonshine [19:39] sounds like beer will only help that discussion [19:39] yeah its pretty tedious [19:39] its like talking to kwmonroe [19:39] hah! [19:40] magicaltrout: if only it were "infrastructure transition to juju implementation" [19:40] now that meeting sounds fun [19:41] sorry, it's 14:40 so i was fetching a beer and missed the context here.. did you need something magicaltrout? [19:42] oh, i see now. hey magicaltrout, maybe bring up "add wifi" in your "infrastructure transition discussion". [19:42] hehe, I tried to get them to go CDK but my sysadmin is a bit of a wimp [19:42] went for docker swarm instead [19:43] docker swarm over kubernetes, interesting [19:43] grim [19:44] although the latest version is much improved [19:44] I bet it was due to the docker registry scanning for vulnerabilities [19:44] I think they have an interesting angle there [19:44] yeah they have put a lot of work into the latest iterations [19:44] I guess would could pair that with k8 though -- although I haven't tried [19:44] so interesting new stuff [19:44] some [19:45] but we do use CDK on our NASA/DARPA project that launched today [19:45] juju on another NASA/DARPA project that has been going for a while now [19:45] so i'm getting it crammed in there ;) [19:46] \o/ [19:50] this is just going through the incubator to TLP vote process kwmonroe https://metron.incubator.apache.org/documentation/ [19:50] might be worth investigating when it gets accepted [19:54] cool [20:07] magicaltrout: i just checked. our namenode uses dfs.permissions.enabled = true, so we're good on big data security ;) [21:59] hmm [21:59] this spark bundle seems to have changed since i last used it [21:59] kwmonroe: know if I can connect to hive these days using beeline? [22:00] I used to be able to do stuff like [22:00] /usr/lib/spark/bin/beeline -u jdbc:hive2://localhost:10000 -e 'CREATE TEMPORARY TABLE store2 USING org.apache.spark.sql.parquet OPTIONS (path \"/user/ubuntu/store.par\")' [22:00] firstly magicaltrout, the spark bundle has indeed changed (you'll want either jujucharms.com/spark-processing/ for spark-standalone or https://jujucharms.com/hadoop-spark/ if you need hdfs and money is no object) [22:01] secondly magicaltrout, where are you serving hive from? did you deploy https://jujucharms.com/hive --to spark? [22:02] s/hive/spark [22:02] just using the hive protocol [22:02] i'm trying to bring an old script we used for a strata demo back to life for arosales but beeline used to provide spark sql support [22:02] aaaahhh, sorry, i didn't grok that at first. and that's not something i generally try to do. what kind of error are you sseing? [22:03] nowt special [22:03] just a connection refused type error [22:04] is the beta controller under high load right now? [22:04] roger that magicaltrout -- could very well be a difference where apache-spark (the old charm) listened on 10000 while the new bigtop-spark does not. lemme deploy and check right quick. [22:04] kwmonroe: issue is the later bigtop spark is on Xenial and saiku with dep on Tomcat is Trusty only [22:05] kwmonroe: so the question here is if the upstream (non-bigtop) apache-spark charm has changed [22:05] oh might be user error [22:05] kwmonroe: specifically the spark used in https://jujucharms.com/u/spicule/saiku-hadoop-spark/ [22:05] hold on [22:07] bdx: you may catch more of the beta/ui folks in #juju-gui too [22:09] scratch that kwmonroe i found the hack i used to have :) [22:11] arosales: thx [22:12] np bdx [22:13] +1 magicaltrout. fwiw magicaltrout arosales, the diffs between that bundle (a-s-6 and latest apache-spark-10) are the inclusion of benchmarks and peer (aka HA) support. nothing related to hivey things between those 2. [22:13] kwmonroe: Thanks [22:16] also magicaltrout, now that arosales has reminded me that you're using trusty subordinates, do *not* use any of the bundles i listed earlier. those are bigtop-based and xenial-only. [22:17] kwmonroe: I would like to see if we can update tomcat for xenial [22:19] arosales: is the saiku sub on xenial? or were you just saying update tomcat in general? [22:19] Saiku is a sub on tomcat as well as spark [22:20] so without getting tomcat to xenial Saiku is stuck on trusty [22:22] saiku isn't or doesn't have to be subordinate to spark I don't think [22:29] magicaltrout: but I think the charm itself is a sub, https://api.jujucharms.com/charmstore/v5/~spicule/trusty/saikuanalytics-enterprise-1/archive/metadata.yaml [22:30] well [22:30] saiku is subordinate to tomcat [22:30] thats for sure [22:30] but there's no technical reason i know of to co locate it on the spark server [22:32] gotcha