[00:26] thumper: huh, really is intermittent for me. i also get TestRemoveMachine failures [02:04] hah, love some of the dummy data in the tests: "pew.pew": "zap", [02:19] veebers: Needs moar "pow!" and "thwok!" [02:19] hah indeed! [02:51] wallyworld: I'm in our HO early if you want to start early [02:51] righto [03:13] wallyworld, thumper seems the queue of jobs is due to goodra being offline, just trying to access it now to see what the haps is [03:13] thansk [03:36] veebers: jump in standup HO when ready? [03:37] wallyworld: yep omw [04:14] veebers: sorry, too quick on clicking mouse [04:14] lol no worries [04:35] huh, turns out I could be using fixup instead of squash when rebasing my branches to save time [04:59] yeah fixup's the bomb [05:19] wallyworld: hurry... there are already conflicts... [05:19] how's it going by the way? [05:19] been doing code review etc after our call, so back onto it now [05:19] trying to multitask [05:20] will have it done today [05:20] :) [05:39] kelvinliu_, wallyworld oh we should re-visit us filling up spaces in aws with our testing and needing to clean it up (kelvin you linked me to something that I lost the tab of, that looked useful) [05:42] veebers, u mean awscli? [05:46] kelvinliu_: it was tags in the vpc we were talking about, I got mixed up [05:47] veebers, yeah, the tags on subnets [05:48] kelvinliu_: I might have just hit a related but different issue, complaining about FAN on spaces [05:48] I'll revisit tomorrow morning, I need to go sort dinner for the fam now [05:49] veebers, ok, enjoy the dinner [05:49] you too o/ see y'all tomorrow [06:04] thumper: fixed it, just tidying up [06:38] cory_fu: A non-reactive charm is being upgraded to a reactive charm using a RelationBase relation per https://github.com/cmars/nrpe-external-master-interface/blob/master/provides.py [06:39] cory_fu: But none of the RelationBase hooks get triggered, so the nrpe-external-master.available flag doesn't get set [06:40] cory_fu: I'm working out a work around, as we likely don't want to fix it for the legacy RelationBase? I don't know what happens with Endpoints. [07:24] thumper: https://pastebin.ubuntu.com/p/mSbw5HtWMp/ [12:37] guild: https://github.com/juju/mgopurge/pull/26 is an update to mgopurge that just brings in our updated dependencies. [12:47] jam: Was just looking at that. Looks like merging #25 gave that one conflicts. [12:48] manadart: yeah, it will, but it should be easy to work out, because 26 should be a superset [12:48] in the end [12:50] jam: Ack. Like you, I couldn't see anything outside of vendor in the diff, but I'll pull it down and see if I can build it. [12:59] manadart: it now has a small patch in main, because juju/txn changed a uint64 to an int [13:43] stub: Yeah, @hooks not firing for relations for non-reactive -> reactive upgrades is a known issue and a big part of why I strongly recommend avoiding @hook whenever possible [13:43] stub: The Endpoint implementations don't depend on @hook so they should be upgrade cleanly [13:44] I think I need to redo that interface as an endpoint since it is so commonly used in my area [13:46] stub: I've generally been converting interface layers whenever I get the chance, as it makes them easier to understand and less error prone. The conversion usually isn't difficult [13:48] stub: Any time scope is part of the exposed API, I replace it with a (relation_id, unit_name) tuple and otherwise keep the API identical, unless the Endpoint pattern suggests a clear improvement that can be done in a backwards-compatible way [13:57] stickupkid: rick: perhaps we should have just one lxd-profile charm for dev/testing instead of an lxd-profile and an lxd-profile-subordinate. [13:58] comment out the blacklist parts in the yaml [13:58] no single blacklist profile will capture all cases. [13:58] if we leave as a local charm, then the file can easily be changed for dev/testing [13:58] hml: I'm not following. [13:59] rick_h_: k, chat after mtg? [13:59] hml: to declare a charm a subordinate we have to define it in the metadata.yaml so I'm not sure how we can get one charm to serve both tests [13:59] hml: k, wfm [13:59] rick_h_: stickupkid: one for lxd and one for lxd subordinate… but no specific black list charms. [14:00] put the blacklist items in the profile yaml,but leave commented out until needed [14:17] magicaltrout: should we block this before it gets merged? https://github.com/apache/bigtop/pull/387 -- we'd need to make a case for keeping oozie (it's currently on the block because it doesn't work with hive2: https://issues.apache.org/jira/browse/BIGTOP-2986) [14:26] Is there any ways to force AWS constaints to allow the instance types t3.* when creating new machines ? Right now I'm getting an invalid constraint value error. [14:28] rick_h_: really enjoyed the juju show, some good stuff in there :) [14:30] stickupkid: yea, I need to get back into making time for those. I think they're good stuff. [14:30] manadart: hmmm, no. Do we not support t3's? [14:31] rick_h_: sec. [14:31] oh sorry, that was tab complete fail manadart [14:31] maaudet: was what I meant to type [14:32] maaudet rick_h_: The instance types recognised from AWS are here: https://github.com/dustinkirkland/instance-type/blob/master/yaml/aws.yaml [14:33] Support of the particular aliases is a LXD concern, we just pass it through. [14:35] Ok, I'll create a pull request [14:36] kwmonroe: yeah i replied in another ticket somewhere about it and said you could go ahead and drop it and we'd repackage it later in the year when we had some cycles to work on it. [14:36] That said, oozie does seem to support Hive2 so i'd like to put the effort in to getting it migrated [14:37] and we could/should update it to 5.0 at the same time anyway [14:39] hml: can I get a CR for this one https://github.com/juju/juju/pull/9142 ? [14:39] stickupkid: sure [14:39] hml: let me know if you need more QA steps, I've made a demo as well, so you can see it in action :D [14:39] stickupkid: nice [14:40] also you owe me an update on the spark pr kwmonroe [14:40] hml: i like making the demos, it forces me to make sure it works :D [14:40] :-) [14:41] magicaltrout: i have a reply in draft. please continue to hold. [14:41] https://www.youtube.com/watch?v=6g4dkBF5anU [14:46] oh yaaas magicaltrout! much <3. i'm gonna use this a lot. [14:47] man those comments are gold... "No point of this video, if you want to hear an hour of this music just call Comcast tech support" [14:48] heheh [14:54] stickupkid: i pushed a non reactive charm to the PR: https://github.com/juju/juju/pull/9141/files… [14:54] stickupkid: checkout the profile yaml, as an example of my proposal above [14:55] hml: nice, i'll take a look [15:05] hml: i thought we where just having key/values in the lxd-profile, I though juju provided the whitelist ? [15:06] s/though/thought/ [15:06] stickupkid: that’s correct [15:06] hml: so what's the blacklist config comments in the profile? [15:07] hml: is that just a reminder to whom is writing it? [15:07] stickupkid: in lue of creating a separate blacklist charm. [15:07] ah [15:07] hml: i think just make a new charm, that way we don't need to edit it? [15:08] stickupkid: there’s a snag with that… we’d need a bunch to catch all the black list items possible. [15:08] the charm doesn’ tneed to be built [15:08] so we can edit the profile.yaml as we’re testing [15:08] it doesn’ tneed to be commited unless we find a bug in it [15:08] hml: i.e. well I was wondering we could just have a charm that just failed, so you can test it manually [15:08] and we know it always fails, then like you said, we can do what you've got there [15:08] i'd be happy with that? [15:09] stickupkid: hrm… could open up issues. the 3 of us chat when rick_h_ get’s back? :-) [15:09] hml sure [15:10] hml: but i like the layout of the lxd-profile :) [15:10] stickupkid: i hope all those work, i was trying to add things from the spec that were specially mentioned [15:10] and i found a few we might want to add to the do not use [15:11] "security.privileged: "true"" <- that one requires a restart [15:11] that'll be interesting :D [15:11] cool [15:12] maybe we should give feedback about which ones require a restart - add it as a nice to have down the road... [15:42] hml: I've updated the PR, i'm going to do some manual testing, to make sure it works, for LXD 2 :) [15:42] stickupkid: ack [15:44] hml: the error message when entering an empty trust-password is really fruity [15:44] hml: "ERROR finalizing credential: missing or empty "client-cert" attribute not valid" [15:45] stickupkid: is an empty trust password valid? [15:45] hml: about to ask the LXD guys [15:52] stickupkid: i have a 1 liner with comments if you have a sec. https://github.com/juju/juju/pull/9144 [15:52] hml: nice and easy one [15:52] yay [15:54] stickupkid: ty! [17:45] stickupkid: hml sorry reading backscroll [17:49] rick_h_: np - not sure stickupkid is around, but we can chat in the am (our) if not [17:50] hml: k, did you want to sync? [17:53] rick_h_: let’s wait for stickupkid, not a rush for today [17:53] hml: ok [20:32] Morning all o/ [20:33] howdy veebers, heads up. I saw that we got the +1 for the 2.4.3 from solutions today [20:35] rick_h_: ack, yay :-) [20:52] kwmonroe: if i wanted to look at the oozie issue [20:52] do i just build the master branch? [21:34] magicaltrout: you mean from a bigtop perspective (you want to see how bigtop fails)? or from upstream oozie? [21:36] i'm mailing the mailing list kwmonroe cause bigtop is making me sad [21:40] magicaltrout: well, if you wanted to try to repro what bigtop sees for oozie, you can use the prebuilt docker images (bottom of https://cwiki.apache.org/confluence/display/BIGTOP/How+to+build+Bigtop-trunk), and "docker run blah ./gradlew bigtop-oozie-pkg" [21:41] er, s/bigtop-oozie-pkg/oozie-pkg/ [21:42] also magicaltrout, if you wanted to build an oozie snap, you can reference the spark snapcraft.yaml... just gonna leave this here: https://github.com/juju-solutions/snap-spark/blob/master/snap/snapcraft.yaml [21:46] i have enough work on... i hate you kwmonroe !.... [21:47] * magicaltrout goes to dig out an espresso [21:48] i also told Ryan you complained about his missing icons.... he swore a bit then complained about inkscape so I believe he's fixing them slowly [21:56] magicaltrout: :) [21:57] magicaltrout: sudo snap install inkscape [21:58] indeed [21:58] that's the hard part.. the rest is just making lines and colors. [21:58] he said it had too many buttons....... [22:01] magicaltrout: sounds like it's time to start a design and UX team in the company :P [22:03] hell if Ryan starts generating some cash [22:03] he can have whatever team he so desires... [22:04] https://www.youtube.com/watch?v=ovZ9iAkQllo this went past my tweetstream a couple of hours ago [22:04] we'll be trying to replicate this in Juju soon [22:05] rick_h_: we're right in thinking you cant sent a mimumum constraint at charm level right, only bundle? We stuck in some code to warn users when their machines were too small today [22:05] but it feels a little clunky [22:07] magicaltrout: I'm trying to parse that atm [22:07] magicaltrout: so yea, the charm can go into an error/etc I guess during install if it fails some sanity checks [22:07] yeah magicaltrout, afaik, constraints can be set in bundles, but not, for example, in a charm's metadata.yaml [22:08] @hook(start): if `free -h < 8GB`, echo "dummy" && exit 1; fi [22:08] thats basically it [22:09] ah look at that Oozie test failures [22:09] winning [22:11] kwmonroe: whats up with your attic, have you moved house or something? [22:24] thumper: https://pastebin.ubuntu.com/p/Y6yX4DdvrQ/ [22:24] oh wait [22:24] wrong package [22:25] doh [22:25] wallyworld: when you have time, we can HO to discuss [22:25] let me check the correct package [22:25] and now i see the failure [22:26] i'll try reverting the changes [22:26] thumper: yup, just revert the changes [22:31] * thumper nods [22:39] hmm, I wonder if my 'hanging' mouse is batteries and not system related :-P Occam's razor and all that [22:40] veebers: yeah, that always gets me too [22:47] veebers, I use a mouse that has a cable that supplies it with power [22:47] externalreality: hah but then you have clutter and, uh, can't take your mouse away from your computer very easily :-P [22:48] for mouse walks [22:48] veebers, touché [23:00] wallyworld: it landed \o/ [23:00] i saw :-) [23:02] magicaltrout: it's the attic in my garage, which i was all gung-ho into converting into an office like 2 years ago, and then i started working on bigtop. [23:02] so i still have to do carpet and sheetrock the knee-walls [23:17] babbageclunk: veebers is crying because he misses you [23:23] it looks rather too clean kwmonroe [23:23] where did the golf clubs and stuff go? [23:33] alright kwmonroe i updated that ticket with some stuff [23:34] chime in BIGTOP-3073 as well if you have an opinion. And reply to my spark PR :P [23:34] bloody kubernetes nonsense