menn0 | jw4: i will keep digging now | 00:00 |
---|---|---|
wallyworld | ericsnow: why the dial changes? | 00:00 |
jw4 | menn0: I'll be in and out for the next few hours... making supper for kids, and then school open house - I'll try to get on longer later | 00:00 |
menn0 | jw4: no worries | 00:00 |
ericsnow | wallyworld: in 1.2 http.Transport doesn't have the KeepAlive field, so I had to manually create the same effect | 00:04 |
wallyworld | ok | 00:05 |
ericsnow | wallyworld: thanks for the review :) | 00:09 |
wallyworld | np | 00:09 |
mup | Bug #1447853 was opened: Local charms are not added to storage on upgrade to 1.22.x <charms> <regression> <storage> <upgrade-juju> <juju-core:Triaged> <https://launchpad.net/bugs/1447853> | 00:19 |
wallyworld | menn0: fixes pushed when you get a chance | 00:23 |
menn0 | wallyworld: kk | 00:23 |
menn0 | wallyworld: ship it | 00:26 |
wallyworld | \o/ | 00:26 |
wallyworld | ty | 00:26 |
mup | Bug #1447853 changed: Local charms are not added to storage on upgrade to 1.22.x <charms> <regression> <storage> <upgrade-juju> <juju-core:Triaged> <https://launchpad.net/bugs/1447853> | 00:31 |
mup | Bug #1447853 was opened: Local charms are not added to storage on upgrade to 1.22.x <charms> <regression> <storage> <upgrade-juju> <juju-core:Triaged> <https://launchpad.net/bugs/1447853> | 00:40 |
sinzui | wallyworld, I am exhausted. I haven't found much. I split bug 1447846 from the other upgrade issue and captured some unit log | 00:52 |
mup | Bug #1447846: Hooks don't fire after upgrade 1.23.0 <hooks> <regression> <upgrade-juju> <juju-core:Triaged by menno.smits> <juju-core 1.23:In Progress by menno.smits> <https://launchpad.net/bugs/1447846> | 00:52 |
wallyworld | sinzui: ty, menn0 is working that bug at the moment | 00:53 |
sinzui | wallyworld, mgz and I also discovered bug 1447853, but I know have a cheep work around | 00:53 |
mup | Bug #1447853: Local charms are not added to storage on upgrade to 1.22.x <charms> <regression> <storage> <upgrade-juju> <juju-core:Triaged> <https://launchpad.net/bugs/1447853> | 00:53 |
wallyworld | ah bollocks, saw that float past | 00:53 |
wallyworld | will have to fix for 1.24 | 00:54 |
sinzui | wallyworld, the work around is to upgrade-charm --force. it works a minute later :) | 00:54 |
wallyworld | at least there is that | 00:54 |
=== kadams54-away is now known as kadams54 | ||
ericsnow | anyone have a minute to review a patch for a very mechanical change: https://github.com/juju/govmomi/pull/2 | 02:01 |
menn0 | fwereade: ping? | 02:30 |
menn0 | jam: ping? | 02:34 |
menn0 | wallyworld: i have a learned a lot more about the hook not firing bug | 03:00 |
menn0 | wallyworld: not sure I know how to fix it though | 03:00 |
wallyworld | oh? | 03:00 |
wallyworld | did you want to talk? | 03:01 |
menn0 | wallyworld: having jam or fwereade around would be handy. it's leader related. | 03:01 |
wallyworld | oh joy | 03:01 |
wallyworld | jam will be online soonish | 03:01 |
menn0 | wallyworld: quick chat would be good | 03:01 |
wallyworld | ok, same hangout | 03:01 |
menn0 | wallyworld: i'm there now | 03:02 |
=== kadams54 is now known as kadams54-away | ||
wallyworld | menn0: if you get a chance, but doesn't matter if not http://reviews.vapour.ws/r/1481/ | 03:25 |
menn0 | wallyworld: got it | 03:26 |
jw4 | menn0: does that mean it's not related to my changes? | 03:40 |
menn0 | jw4: at this stage it's looking like the hooks not firing issue is related to the leadership work | 03:41 |
menn0 | jw4: i'm not any wiser about that mode error in the logs though | 03:41 |
jw4 | menn0: well I know about the error - that's not critical | 03:42 |
jw4 | menn0: cool, I'll try and dig into it too | 03:42 |
jw4 | for my edification | 03:42 |
menn0 | we just had an earthquake, 6.4 but very deep | 03:43 |
jw4 | whoa | 03:44 |
menn0 | because of the depth it wasn't that severe | 03:44 |
jw4 | menn0: any damage | 03:44 |
menn0 | but my chair was rolling around by itself | 03:44 |
menn0 | no damage i think but it was a strange rolling feeling and went on for a while | 03:45 |
menn0 | i've only experienced short "bangs" up until now | 03:45 |
anastasiamac | menn0: is south island sinking? | 03:45 |
jw4 | that's freaky... are earthquakes common in that area? Pacific Rim and all | 03:45 |
menn0 | yep, quite common in NZ | 03:45 |
jw4 | anastasiamac: he's ignoring the question... I think he's rescuing pets or something | 03:48 |
menn0 | anastasiamac: probably :) | 03:49 |
jw4 | quite a bit of activity recently: http://www.geonet.org.nz/quakes/felt | 03:50 |
anastasiamac | jw4: :D | 03:51 |
jw4 | :) | 03:51 |
anastasiamac | menn0: u should have apost-earthquake party, considering it's friday | 03:51 |
menn0 | anastasiamac: they're a little too common to be having parties for each one :) | 03:56 |
anastasiamac | menn0: on a serious note, it's gr8 that it was deep. 6.4 sounds really scary... :)) | 03:57 |
menn0 | anastasiamac: yes. this one was the same magnitude as the one that did this a few years ago: http://en.wikipedia.org/wiki/2011_Christchurch_earthquake | 03:59 |
menn0 | the city is still a long way from recovering from that one | 04:00 |
wallyworld | menn0: the earth moved for you in cape town as well didn't it? | 04:09 |
menn0 | wallyworld: ssh not in public :) | 04:10 |
mup | Bug #1447895 was opened: Panic if jujud restarts while action is running <actions> <juju-core:New> <https://launchpad.net/bugs/1447895> | 04:11 |
mup | Bug #1447899 was opened: upgrade fails if no explicit version is specified <upgrade-juju> <juju-core:Triaged> <https://launchpad.net/bugs/1447899> | 04:11 |
jw4 | nice sleuthing menn0 | 04:28 |
menn0 | jw4: with the hook not firing bug? | 04:28 |
jw4 | yeah | 04:28 |
jw4 | finding the WatchLeadershipSettings missing document | 04:29 |
menn0 | wallyworld, jw4: bingo! | 04:50 |
menn0 | wallyworld, jw4: I have a fix for the hook not firing issue | 04:50 |
menn0 | wallyworld, jw4: adding an upgrade step that adds the leadership documents in the settings collection fixes the issue | 04:51 |
menn0 | wallyworld, jw4: I hacked it in without tests so let me sort that out before proposing | 04:51 |
jw4 | menn0: suh-weet! | 04:52 |
menn0 | wallyworld, jw4: initial, rough fix for hooks not firing issue | 04:59 |
menn0 | https://github.com/mjs/juju/tree/1447846-hooks-dont-fire-after-upgrade-1.23 | 04:59 |
menn0 | writing the tests now | 04:59 |
jw4 | other than missing docstring on https://github.com/juju/juju/compare/1.23...mjs:1447846-hooks-dont-fire-after-upgrade-1.23#diff-d7b2b2c8e8ce6dfc1b7f09c3cf9744d1R1203 | 05:04 |
jw4 | and forthcoming tests... | 05:04 |
jw4 | looks great menn0 | 05:04 |
wallyworld | menn0: branch looks good, still want to understand why listener hangs if there's no record. i think that's poor behaviour | 05:13 |
jw4 | wallyworld: I bet it's because the watcher isn't returning an initial empty event, like it's supposed to | 05:14 |
jw4 | wallyworld, menn0 it was a little tricky getting that initial guaranteed event in the Action Watchers, so I wouldn't be surprised if the Leadership Watchers have a problem there too | 05:15 |
menn0 | jw4: I think the problem here is that the code assumes the leadership doc will be there b/c it is for services created under 1.23 | 05:15 |
menn0 | jw4: but someone forgot to write this upgrade step | 05:16 |
jw4 | the Watchers (afair) are supposed to guarantee an initial event, possibly empty, even if there are no records... | 05:16 |
menn0 | jw4: and without the doc there when the watcher is created the initial event isn't fired | 05:16 |
menn0 | jw4: are you sure about that? | 05:16 |
jw4 | menn0: yeah, I think you're right, but I remember fwereade stressing to me the importance of that... (but I could be remembering wrong) | 05:16 |
jw4 | (the importance of the initial, possibly empty, guaranteed event) | 05:17 |
menn0 | jw4: yeah, I know there's been talk about this but I'm not exactly sure what should happen | 05:17 |
jw4 | hopefully fwereade will be on soon and will chime in... I guess it's EOD for you though now (and EOW) | 05:18 |
menn0 | jw4: i've got a bit more time left | 05:19 |
menn0 | jw4: i'm going to try and get this fix done as much as I can | 05:19 |
menn0 | jw4: regardless of what the watchers are supposed to do I'm pretty sure we want this upgrade step | 05:20 |
jw4 | menn0: regardless, I think your fix is appropriate in this case | 05:20 |
wallyworld | jw4: exactly, but to me that's not how i thought the watchers worked | 05:20 |
wallyworld | +1 | 05:21 |
jw4 | wallyworld: yeah, I'm fairly confident that's how I was instructed to implement the actions related watchers, but we should verify | 05:21 |
* jw4 off to bed | 05:32 | |
=== urulama__ is now known as urulama | ||
menn0 | wallyworld: ping | 05:51 |
wallyworld | yo | 05:52 |
menn0 | wallyworld: i have to EOD very soon | 05:52 |
wallyworld | one sec | 05:52 |
wallyworld | back | 05:54 |
menn0 | wallyworld: just need to hand over the fix for this | 05:54 |
menn0 | wallyworld: pushing now... | 05:54 |
menn0 | wallyworld: give me a sec | 05:54 |
wallyworld | sure, is it ready to land after review? | 05:54 |
wallyworld | or tests needed? | 05:54 |
menn0 | wallyworld: the hard tests are done | 05:55 |
menn0 | wallyworld: just needs a test for upgrade step idempotency | 05:55 |
wallyworld | ok | 05:56 |
menn0 | it's here: https://github.com/mjs/juju/tree/1447846-hooks-dont-fire-after-upgrade-1.23 | 05:58 |
wallyworld | menn0: i'll either do it or hand over; it's my son's birthday today and i have to go to dinner soonish. either way will be done. thanks for fixing | 05:58 |
menn0 | it's ready to go apart from that one test | 05:58 |
menn0 | if no-one gets it done today it'll take me almost no time on Monday | 05:59 |
menn0 | wallyworld: and we need to forward port to master | 05:59 |
wallyworld | ok, np. tbh, i don't think we are cutting a release on friday anyway | 05:59 |
wallyworld | but it's now done so the risk is removed | 06:00 |
menn0 | wallyworld: ok good | 06:02 |
menn0 | wallyworld: well have a good night and a weekend | 06:02 |
menn0 | speak to you next week | 06:02 |
wallyworld | fwereade: you online? | 06:54 |
wallyworld | fwereade: hiya | 10:17 |
wallyworld | TheMue: hi there, you ocr? would love a review of this so as it needs to land for 1.24 http://reviews.vapour.ws/r/1481/ it removes a feature flag | 10:28 |
wwitzel3 | jam: ping | 11:13 |
mattyw | wwitzel3, ping? | 11:27 |
wwitzel3 | mattyw: pong | 11:30 |
fwereade | wallyworld_, so do we still serve client facade v0? I'm not seeing where we do | 11:40 |
fwereade | wallyworld_, and that would STM to be a problem... | 11:40 |
wallyworld_ | fwereade: ah, right, i wasn't sure about that. i wasn't sure what the exact mechanism was there | 11:41 |
fwereade | wallyworld_, the ideal, I think, is to *never touch* the client API, and carve off changing functionality into more service-specific facades | 11:42 |
fwereade | wallyworld_, otherwise you have to have an almost-perfectly-cloned Client facade | 11:42 |
fwereade | wallyworld_, then when (say) storage evolves, it shouldn't need to hit more than one or two small facades | 11:43 |
fwereade | wallyworld_, rather than dragging the Client mess into the equation every time | 11:43 |
fwereade | wallyworld_, what actually changed on Client? | 11:44 |
wallyworld_ | fwereade: yeah sadly i think you are right, i'll have to do that, but will miss the branching of 1.24. bollocks. will have to forward port | 11:44 |
fwereade | wallyworld_, bad luck :( | 11:44 |
wallyworld_ | fwereade: new params to 2 apis were added | 11:44 |
wallyworld_ | fwereade: if a newer client adds additonal params to the call, they will be ignored, but we need a way to tell the user | 11:45 |
wallyworld_ | hence the api bump | 11:45 |
fwereade | wallyworld_, yeah, indeed, it really is a different API | 11:45 |
fwereade | wallyworld_, so I *think* that the right thing to do | 11:46 |
fwereade | wallyworld_, is to pull out a Service service facade from client | 11:46 |
wallyworld_ | yeah, i suspect so too | 11:46 |
wallyworld_ | i'll revisit over the weekend | 11:46 |
wallyworld_ | i really want to remove the flag for 1.24 | 11:47 |
perrito666 | morning | 12:35 |
mgz | perrito666: wala | 12:44 |
redelmann | perrito666, ¿good? morning | 12:45 |
=== kadams54 is now known as kadams54-away | ||
=== kadams54-away is now known as kadams54 | ||
katco | wwitzel3: ericsnow: stand up | 14:02 |
=== kadams54 is now known as kadams54-away | ||
=== kadams54-away is now known as kadams54 | ||
=== kadams54 is now known as kadams54-away | ||
=== kadams54-away is now known as kadams54 | ||
=== liam_ is now known as Guest67926 | ||
alexisb | all, looking at this: https://bugs.launchpad.net/juju-core/+bug/1447846 | 15:56 |
mup | Bug #1447846: Hooks don't fire after upgrade 1.23.0 <hooks> <regression> <upgrade-juju> <juju-core:Triaged by menno.smits> <juju-core 1.23:In Progress by menno.smits> <https://launchpad.net/bugs/1447846> | 15:56 |
alexisb | ^^^ based on menno's last comment is this something someone can pick up and complete today? | 15:57 |
jw4 | alexisb: menn0 said it's almost done.. just one test for idempotency left | 15:57 |
katco | alexisb: yeah looks doable for someone to pick up | 15:58 |
alexisb | katco, can you please delegate | 15:58 |
katco | alexisb: sure | 15:58 |
alexisb | thanks | 15:58 |
katco | moonstone are all currently working on bugs. any volunteers? | 15:59 |
katco | cherylj: perrito666: do you two have capacity to land a simple bug? | 16:03 |
=== urulama is now known as urulama__ | ||
perrito666 | katco: sorry I dont feel really reliable right now. | 16:07 |
katco | perrito666: no worries. feel better soon | 17:04 |
katco | cherylj: sorry i lost connection. can you take that bug? | 17:04 |
natefinch | man I really hate that we require go 1.2 for juju, it totally screws up my development environment for everything else... and many projects already require 1.3+ | 17:23 |
katco | natefinch: here's my setup | 17:26 |
katco | natefinch: ~/.local/go-1.{2,3,4} | 17:26 |
katco | natefinch: ~/.local/go -> go-1.2 during day | 17:26 |
katco | natefinch: export GOPATH=~/.local/go | 17:26 |
katco | natefinch: there is some churn with having to rebuild utils | 17:26 |
ericsnow | hmmm...virtualenv for go... | 17:27 |
katco | ericsnow: or there's that :) | 17:27 |
natefinch | ericsnow: there's a tool for that... forget which of the myriad tools out there it is. of course, as katco said, you can always just run them side by side. | 17:29 |
katco | natefinch: wwitzel3: planning time! | 17:32 |
wwitzel3 | I'm there? | 17:32 |
wwitzel3 | is it not in moonstone? .. (checks calendar) | 17:33 |
natefinch | wwitzel3: not moonstone | 17:34 |
=== kadams54 is now known as kadams54-away | ||
=== kadams54-away is now known as kadams54 | ||
wwitzel3 | cmars: http://reviews.vapour.ws/r/1484/ | 17:50 |
cmars | wwitzel3, so i guess CI is going to set that feature flag to get that test to pass? what's the difference between not registering the provider vs. blocking bootstrap? | 17:55 |
wwitzel3 | cmars: the code on the unit doesn't see the provider as registered since the flag is local | 17:57 |
wwitzel3 | cmars: jam pointed this out no the ML and the QA ran in to it with CI tests | 17:57 |
aznashwan | hey; could I please get a quick review on http://reviews.vapour.ws/r/1486/ | 17:58 |
cmars | wwitzel3, ok, got it. so this moves the featureflag check to the client? in what context does environ.Bootstrap run? | 17:58 |
cmars | (environ's not an area of the codebase I'm familiar with) | 17:58 |
aznashwan | it's a simple module; ~100 lines of code that actually does something | 17:58 |
wwitzel3 | cmars: just during juju bootstrap | 18:03 |
=== kadams54 is now known as kadams54-away | ||
natefinch | cmars: quick review? http://reviews.vapour.ws/r/1487/diff/# | 18:34 |
wwitzel3 | cmars: thanks for the review | 18:40 |
perrito666 | /query alexisb | 19:53 |
perrito666 | well, another bug in my irc client | 19:53 |
perrito666 | I should stop using daily snapshots | 19:53 |
natefinch | sometimes I think Juju is the biggest example of what not to do with interfaces :/ | 20:24 |
perrito666 | you exagerate | 20:25 |
natefinch | perhaps slightly | 20:25 |
natefinch | there might be a few worse examples in the world ;) | 20:25 |
natefinch | I like the way everything is a nice generic interface and then in the comments we just say "this is for precise0 | 20:27 |
natefinch | s/0/" | 20:27 |
perrito666 | natefinch: I think that is an example on how not to used a typed language | 20:34 |
mgz | anyone still around for trivial reviews? | 21:10 |
mup | Bug #1448308 was opened: Skipped TestUniterUpgradeConflicts on ppc64 <skipped-test> <test-failure> <juju-core:Triaged> <https://launchpad.net/bugs/1448308> | 21:16 |
ericsnow | mgz: sorry I missed that format stuff :/ | 21:21 |
ericsnow | anyone around that feels comfortable reviewing http://reviews.vapour.ws/r/1490/ | 21:22 |
ericsnow | it fixes the "hooks not firing" bug | 21:23 |
ericsnow | #1447846 | 21:23 |
mup | Bug #1447846: Hooks don't fire after upgrade 1.23.0 <hooks> <regression> <upgrade-juju> <juju-core:Triaged by menno.smits> <juju-core 1.23:In Progress by ericsnowcurrently> <https://launchpad.net/bugs/1447846> | 21:23 |
mgz | ericsnow: I can look | 21:23 |
ericsnow | mgz: oh, cool | 21:23 |
mgz | ericsnow: I also have a trivial ppc64 test skip branch | 21:24 |
ericsnow | mgz: note that I only added the idempotency test, menn0 did the rest | 21:24 |
mgz | /1488 | 21:24 |
ericsnow | mgz: I'll take a look | 21:24 |
mgz | ericsnow: yeah, I already looked over menno's branch | 21:24 |
ericsnow | mgz: nice | 21:24 |
ericsnow | mgz: I'm pretty sure I got the idempotency check right | 21:25 |
ericsnow | mgz: in the bug you make it sound like skipping the test might be masking a real bug | 21:27 |
mgz | ericsnow: it might be, but we hav ehte bug, and know we need to do functional testing of the leadership stuff | 21:28 |
ericsnow | mgz: k | 21:28 |
ericsnow | mgz: so 1448308 will be closed only as soon as we resolve the issue (and not when the skip-the-test patch merges), right? | 21:29 |
mgz | ericsnow: how-upgrades-work question | 21:29 |
ericsnow | mgz: sure | 21:29 |
mgz | ericsnow: indeed, I'm keeping the skipped-tests bugs open, because we keep filing bugs and linking them in the codebase, but counting them as fixed when the test is skipped | 21:30 |
mgz | which is wrong... | 21:30 |
ericsnow | mgz: cool, then LGTM | 21:31 |
mgz | I'm on 1.23.0 which did not do this step, my state server notices there's a 1.23.2 with this fix and downloads it... how does it know to run this step? | 21:31 |
mgz | we just always do steps on new versions and the steps must apply? | 21:32 |
mgz | even minor ones, right? | 21:32 |
ericsnow | mgz: I'm not sure, but...the idempotency invariant of upgrade steps ensures it's a noop if you're already okay | 21:33 |
mgz | right | 21:33 |
ericsnow | I expect it simply runs all the steps | 21:34 |
mgz | ericsnow: test changes look good | 21:37 |
ericsnow | mgz: k, cool | 21:37 |
ericsnow | mgz: it made sense to factor out that stuff into helpers :) | 21:38 |
mgz | I'd be somewhat interested in a test that ran the step on a as-if just bootstrapped system, no services | 21:38 |
mgz | we don't try and do all error paths in these things, but that one does seem relevent | 21:39 |
ericsnow | mgz: so basically the same thing as the existing test but without adding any services first? | 21:41 |
mgz | (it should just return success without running any ops basically) | 21:42 |
mgz | ericsnow: yup | 21:42 |
ericsnow | mgz: I'll add such a test! | 21:42 |
mgz | ericsnow: lgtm | 21:42 |
mgz | hm, can I shipit and leave a comment in the same thing? not sure on the magic | 21:43 |
mgz | I guess I jsut include that string in my text huh. | 21:43 |
ericsnow | mgz: so is it correct that CI covers the upgrade scenarios now (i.e. the bug trigger will get exercised)? | 21:58 |
mgz | ericsnow: see all the failures on upgrade jobs on 1.23 and trunk :) | 21:59 |
ericsnow | mgz: that's what I figured :) | 22:00 |
ericsnow | mgz: thanks for the review | 22:00 |
mgz | http://reports.vapour.ws/releases/2557 | 22:00 |
mgz | "panic: rescanned document misses transaction in queue" ...that one's new to me | 22:30 |
=== kadams54-away is now known as kadams54 | ||
ericsnow | mgz: yeah, me too :\ | 22:43 |
ericsnow | mgz: hmmm...see #1318366 | 22:47 |
mup | Bug #1318366: jujud on state server panic misses transaction in queue <cloud-installer> <landscape> <orange-box> <panic> <performance> <sm15k> <juju-core:Fix Released by menno.smits> <juju-core 1.20:Fix Released by menno.smits> <juju-core (Ubuntu):Fix Released> <https://launchpad.net/bugs/1318366> | 22:47 |
fwereade | ericsnow, menn0, dammit, sorry, thanks | 22:48 |
ericsnow | mgz: I'm going to ask menn0 to double check the upgrade step and tests when he gets a chance; that failure is a little suspicious. | 23:00 |
mgz | ericsnow: seems wise, we're not in a rush now | 23:06 |
ericsnow | mgz: yep :) | 23:07 |
=== kadams54 is now known as kadams54-away |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!