* redir wonders where — and – fall in that spectrum... | 00:14 | |
redir | they look so similar | 00:14 |
---|---|---|
redir | and how about ‗ ? | 00:16 |
alexisb | babbageclunk, dude! it is time for sleeping | 00:21 |
* alexisb goes to pick up kiddo talk to you all later | 00:21 | |
=== alexisb is now known as alexisb-afk | ||
babbageclunk | alexisb: you are right! I'm all done, except can't do a final test because of some weird dns issue. | 00:22 |
* babbageclunk goes to bed! | 00:22 | |
wallyworld | axw: i'd like to get this landed for beta if possible, it's a small tweak to list controllers http://reviews.vapour.ws/r/5635/ | 01:37 |
axw | looking | 01:45 |
axw | wallyworld: reviewed | 01:57 |
wallyworld | ta | 01:57 |
wallyworld | axw: 1:1? | 02:02 |
=== natefinch-afk is now known as natefinch | ||
natefinch | I hate it when I fix a simple bug and it reveals a huge gnarly bug behind that one. | 02:11 |
anastasiamac_ | natefinch: only one gnarly behind? consider it ur lucky day :D | 02:14 |
natefinch | well.. it's one gnarly bug because of two pieces of gnarly code working at cross purposes... both assuming too much about what the other one will do or not do | 02:15 |
natefinch | axw: got a minute? | 02:17 |
natefinch | axw: I'm going to ask you about code you wrote threee years ago ;) | 02:17 |
natefinch | start dusting off the tape drive :) | 02:18 |
axw | natefinch: sorry, missed your message | 02:49 |
axw | natefinch: what's up? | 02:49 |
natefinch | axw: heh, no problem. I've actually engineered a fix around it, but we can talk a little... | 02:50 |
natefinch | axw: juju help-tool | 02:51 |
natefinch | axw: https://github.com/juju/juju/blame/master/cmd/juju/commands/helptool.go#L123 | 02:51 |
natefinch | axw: we make a dummy hook context and then pass that into all the functions and hope they don't look too closely at it :p | 02:51 |
axw | let me guess, something is looking at it? :) | 02:52 |
natefinch | axw: yes :) | 02:52 |
natefinch | axw: but they don't *really* need to, at least not so early, and since all this code does is call info... I just deferred the code that looks at the value until we actually run the command. | 02:53 |
axw | natefinch: ok. perhaps the interface for those commands should be changed, so that Run is the thing that takes the hook context? | 02:54 |
natefinch | axw: that's a good general solution. I don't think anything should really need to use the hook context on command creation, and "context" is really more of a "while running" concept anyway (which is why we pass in the command context there) | 02:55 |
natefinch | axw: I made the minor change to get this one to work. I think in the future it would be good to make the fix you suggested, but that seems like a bigger change than we really have time for right now. | 03:02 |
natefinch | axw: http://reviews.vapour.ws/r/5636/diff/# | 03:02 |
natefinch | or if anyone else wants an easy review ^ (+26 -12) | 03:10 |
axw | natefinch: reviewed, please add QA steps tho | 03:12 |
axw | (code LGTM) | 03:12 |
natefinch | oops, right, updated | 03:14 |
axw | natefinch: do you know if we have a CI test that uses resource-get? | 03:15 |
natefinch | axw: not sure, I can check | 03:15 |
natefinch | axw: yeah, there is | 03:25 |
axw | natefinch: cool, thanks | 03:25 |
wallyworld | ffs, landing bot is down :-( | 03:49 |
axw | wallyworld: can you please check if you can see the data I just shared on drive? | 04:19 |
wallyworld | axw: yep, am looking at it now. i think it would be good to extract the 3 interesting numbers for each run and tabulate | 04:20 |
axw | wallyworld: indeed, I need to prepare something for my talk later tho :/ I've responded with rough figures that are most interesting, will see what else I can do | 04:21 |
wallyworld | axw: i can pull out some numbers if i get time a bit later | 04:21 |
wallyworld | axw: ah i see the next email; i think that's ok for now, it gives the key data point | 04:22 |
wallyworld | axw: only other data point would be what it was before your changes | 04:23 |
axw | wallyworld: yeah, but I figured it's not that useful to know how bad the really bad implementation is :p | 04:24 |
axw | I'd prefer to focus on how to get it better | 04:24 |
wallyworld | fair enough | 04:25 |
menn0 | axw or wallyworld: could I get a review of this please? http://reviews.vapour.ws/r/5629/ | 04:39 |
menn0 | it's fully QAed now | 04:39 |
wallyworld | ok | 04:39 |
wallyworld | menn0: except it appears landing bot is down :-( | 04:39 |
menn0 | wallyworld: ah... oh well | 04:40 |
menn0 | wallyworld: the review would be good to get done anyway | 04:40 |
wallyworld | sigh | 04:40 |
wallyworld | yep | 04:40 |
* menn0 looks at the bot | 04:40 | |
wallyworld | menn0: oh, it's back | 04:40 |
menn0 | wallyworld: ok cool | 04:41 |
wallyworld | got a server error before | 04:41 |
menn0 | wallyworld: TestCertificateUpdateWorkerUpdatesCertificate is failing really often | 04:48 |
wallyworld | awesome | 04:48 |
wallyworld | it has been ok for a while, was failing a bit a while back | 04:48 |
natefinch | menn0: how is it failing? | 04:48 |
menn0 | natefinch: http://juju-ci.vapour.ws/job/github-merge-juju/9153/artifact/artifacts/windows-out.log | 04:49 |
menn0 | natefinch: the test thinks the certificate isn't being updated | 04:49 |
menn0 | natefinch: it's happened to me twice in a row for a PR that changes a very unrelated shell script | 04:49 |
wallyworld | it should even be running on windows | 04:49 |
wallyworld | shouldn't | 04:49 |
menn0 | it failed for me yesterday a few times too | 04:50 |
wallyworld | anything controller related does not need to run on windows | 04:50 |
menn0 | wallyworld: agreed, but this doesn't seem like a windows specific problem | 04:50 |
menn0 | i.e. maybe we're just lucky that it's not failing more often on linux | 04:50 |
natefinch | [LOG] 0:05.803 DEBUG juju.worker.certupdater addresses haven't really changed since last updated cert | 04:51 |
natefinch | I was looking into this cert updater stuff a lot for a manual provider bug I was looking into, so I'm fairly familiar with the code. There's a watcher here that's getting fired, but we think nothing has changed. If nothing has changed, why is the watcher firing? | 04:52 |
menn0 | natefinch: could be an update to the doc which just set the same values? that would still cause the watcher to fire. | 04:57 |
natefinch | yeah, true. Still, seems suspicious that we're getting that "nothing changed" firing and then the test failed because nothing changed. | 04:57 |
menn0 | natefinch: true | 04:59 |
menn0 | natefinch: I just dug through the code. the watcher is tracking the machine document so any change to the machine will fire it. not all those changes will be address updates. | 05:00 |
natefinch | oh right, that makes sense | 05:00 |
menn0 | natefinch: so that message could be completely normal | 05:00 |
menn0 | or it's a clue :) | 05:01 |
natefinch | is the landing bot working, or not? | 05:02 |
wallyworld | menn0: lgtm | 05:09 |
wallyworld | natefinch: it is again now | 05:09 |
menn0 | wallyworld: thanks | 05:18 |
wallyworld | menn0: i think i found the pinger issue. haven't tested yet, just looking at code | 05:31 |
wallyworld | but i don't fully understand it either | 05:31 |
menn0 | wallyworld: for the cert updater? | 05:33 |
wallyworld | menn0: in apiserver/admin.go there's a call to startPingerIfAgent() | 05:34 |
wallyworld | but it is inside an if{} | 05:34 |
wallyworld | and i think nothing ever calls it | 05:34 |
wallyworld | that's the only place i can see where the pinger would be started | 05:34 |
wallyworld | i'll add some logging and test a bit | 05:35 |
menn0 | wallyworld: I believe the if there to avoid unnecessarily starting pingers for controller machine agents which log in on behalf of a model. | 05:42 |
menn0 | the thinking is there's no need to run the pinger if the client is on the same machine | 05:43 |
wallyworld | ok. i'll confirm if we are starting the pinger or not and gp from there | 05:43 |
menn0 | sounds good | 05:43 |
wallyworld | fwereade: you around? | 07:16 |
fwereade | wallyworld, yeah, sorry | 07:28 |
mup | Bug #1493058 changed: ensure-availability fails on GCE <docteam> <gce-provider> <ha> <jujuqa> <juju:Triaged> <juju-core:Invalid> <juju-core 1.24:Won't Fix> <juju-core 1.25:Won't Fix> <https://launchpad.net/bugs/1493058> | 08:34 |
anastasiamac_ | wallyworld: u've kindly file this bug... but I *think* this has now been fixed as off last week. right? :D | 08:46 |
anastasiamac_ | https://bugs.launchpad.net/juju/+bug/1559701 | 08:46 |
mup | Bug #1559701: kill-controller manual provider broken <2.0-count> <manual-provider> <manual-story> <juju:Triaged> <https://launchpad.net/bugs/1559701> | 08:46 |
anastasiamac_ | axw: do u know? ^^ | 08:46 |
axw | anastasiamac_: that's fixed. destroy-controller auto destroys manual machines now | 08:47 |
anastasiamac_ | axw: \o/ | 08:48 |
dimitern | mgz: hey there - about? | 09:11 |
voidspace | babbageclunk: ping | 09:12 |
babbageclunk | voidspace: pong | 09:13 |
mgz | dimitern: hey | 09:52 |
dimitern | mgz: I'm looking into bug 1621538 | 09:54 |
mup | Bug #1621538: container networking: cannot juju ssh to container <ci> <maas-provider> <netowork> <regression> <juju:In Progress by dimitern> <https://launchpad.net/bugs/1621538> | 09:54 |
dimitern | mgz: it seems the error is different | 09:55 |
dimitern | mgz: not about parent br-eth0 not having an address but something even weirder - node with that [device] system_id already exists | 09:56 |
mgz | dimitern: there are only two changes in the regresion window | 09:59 |
mgz | pr #6156 and pr #6158 | 09:59 |
dimitern | mgz: still digging in both maas and juju sources to see how that duplicate error can even happen - we're namespacing the hostnames(instance ids) we generate for containers and their backing maas devices | 10:00 |
babbageclunk | dimitern: dhcp won't start on my maas :( complaining about a missing conf file | 10:33 |
voidspace | fwereade: ping | 10:34 |
fwereade | voidspace, pong | 10:34 |
dimitern | babbageclunk: is it inside LXD ? | 10:34 |
voidspace | fwereade: so I'm trying to build an operation that "marks the action as failed and restores local state to a sensible state" | 10:34 |
babbageclunk | dimitern: I think it's because the drive filled up - I made the fs bigger. Any way I can get it to rewrite the config file? | 10:34 |
babbageclunk | dimitern: no, kvm | 10:34 |
fwereade | voidspace, ah yes | 10:34 |
voidspace | fwereade: afaics marking an action as failed means calling "state.State.ActionFinished" | 10:35 |
dimitern | babbageclunk: I'd try dpkg-reconfigure maas-dhcp first | 10:35 |
voidspace | fwereade: and nothing in the uniter/actions/resolver has access to that | 10:36 |
voidspace | fwereade: hmm... op_callback has some code that does that | 10:36 |
fwereade | voidspace, I thought I saw a method on callbacks that did that | 10:36 |
voidspace | fwereade: right | 10:36 |
fwereade | voidspace, so you should be able to supply that capability in the op factory | 10:36 |
babbageclunk | dimitern: just tried that, no dice. | 10:36 |
babbageclunk | dimitern: I guess it's reinstall time! | 10:37 |
voidspace | fwereade: right, sounds good to me :-) thanks | 10:37 |
dimitern | babbageclunk: what's the error you're getting? | 10:38 |
fwereade | voidspace, and the "sensible state" would be "that achieved by running the same code as in RunAction.Commit", I think | 10:38 |
babbageclunk | dimitern: http://paste.ubuntu.com/23154038/ | 10:38 |
voidspace | fwereade: yeah, that I'm fine with (well - until I actually get to doing it...) | 10:38 |
babbageclunk | dimitern: And it's right - that file's not there. | 10:38 |
fwereade | voidspace, cool :) | 10:39 |
dimitern | babbageclunk: ok, before reinstalling, try also dpkg-reconfigure maas-rack-controller | 10:39 |
babbageclunk | dimitern: ooh yeah, that looks better! | 10:40 |
babbageclunk | dimitern: didn't seem to help - I'll try bouncing the machine just in case | 10:42 |
dimitern | babbageclunk: yeah - and also check the permissions on /var/lib/maas/* | 10:43 |
babbageclunk | dimitern: Hold the phone, actually I think that did it! Yay thanks! | 10:43 |
dimitern | babbageclunk: \o/ :) | 10:43 |
mup | Bug # changed: 1543660, 1546805, 1587644, 1592887, 1621658 | 11:04 |
mup | Bug # opened: 1543660, 1546805, 1587644, 1592887, 1621658 | 11:10 |
mup | Bug # changed: 1543660, 1546805, 1547806, 1587644, 1592887, 1621658 | 11:16 |
mup | Bug #1547806 opened: open-port does not work on EC2 <juju:Triaged> <https://launchpad.net/bugs/1547806> | 11:28 |
rick_h_ | dimitern: morning, how goes the battle with the functional container network tests? | 11:28 |
dimitern | rick_h_: morning, still trying to repro the new error | 11:29 |
dimitern | rick_h_: not related to missing addresses on host bridges | 11:30 |
rick_h_ | dimitern: rgr | 11:30 |
dimitern | rick_h_: it seems maas 2.0 specific so far | 11:30 |
rick_h_ | dimitern: they just upgraded to maas 2.0 final in CI yesterday before this run | 11:30 |
dimitern | rick_h_: I thought at first that job runs on maas 1.9 and wasted some time trying to find the error in lp:maas/1.9 | 11:31 |
dimitern | rick_h_: good that I checked the job's jenkins config | 11:32 |
rick_h_ | dimitern: it came up yesterday that it was on 2.0rc3 I think and was upgraded to 2.0 final | 11:32 |
dimitern | rick_h_: ok, that's useful - I'll concentrate on changes since 2.0rc3 | 11:33 |
mup | Bug #1547806 changed: open-port does not work on EC2 <juju:Triaged> <https://launchpad.net/bugs/1547806> | 11:34 |
* rick_h_ goes to get the boy to school and make a coffee run, biab | 11:42 | |
fwereade | babbageclunk, http://reviews.vapour.ws/r/5637/ is up if you have a moment; ask many questions if required :) | 11:45 |
frobware | dimitern: ping | 12:28 |
dimitern | frobware: pong | 12:28 |
frobware | dimitern: any joy with the container failure? | 12:28 |
dimitern | frobware: so far it seems something broke after the maas 2.0rc3 on finfolk got upgraded to 2.0 final | 12:29 |
frobware | dimitern: the maas update that we did with mgz yesterday? | 12:30 |
frobware | dimitern: (I thought that 1.9) | 12:30 |
dimitern | frobware: I though so as well, but the job runs on 2.0 | 12:30 |
mgz | that was a different maas | 12:30 |
mgz | and this failure has been happening since the landings on the 7th | 12:31 |
dimitern | mgz: the first time it happend was after frobware's PR got reverted, so it can reach a bit further preparing container NICs | 12:32 |
* frobware jumps on a train; back in 90 mins... or so. | 12:34 | |
babbageclunk | fwereade: was lunching - looking now | 12:37 |
fwereade | babbageclunk, cheers | 12:42 |
natefinch | rick_h_: got a suggestion for another bug to pick up? | 12:48 |
rick_h_ | natefinch: for that one I pushed back on not doing anything on that | 12:49 |
rick_h_ | natefinch: did you get the reply? | 12:49 |
natefinch | rick_h_: missed that, will go look | 12:49 |
rick_h_ | ty | 12:49 |
natefinch | gah, I need multiple return values on my executables :/ | 12:53 |
babbageclunk | GAH, can someone please tell me what I should avoid so that the bot will post to RB? | 12:53 |
natefinch | heh | 12:53 |
babbageclunk | double gah! | 12:53 |
babbageclunk | o/ | 12:53 |
natefinch | babbageclunk: not sure what's wrong with the rb bot | 12:53 |
babbageclunk | Is it working for others at the moment? | 12:55 |
natefinch | babbageclunk: last night it made the review for me, but didn't update my PR description with the link | 12:56 |
babbageclunk | I think it's something to do with things I put in the description - it's looking for (Review request: http://blah) so it knows whether to update or create a new one. But it gets confused. | 12:57 |
rick_h_ | natefinch: what about going with resource-get-fingerprint | 12:57 |
natefinch | rick_h_: well, seems like the information we really want is "is the info different on the server" Do we care what the actual fingerprint is? I don't think that tells us anything by itself | 13:00 |
babbageclunk | dimitern, voidspace, frobware: can I get a review of http://reviews.vapour.ws/r/5638/ please? | 13:00 |
dimitern | babbageclunk: looking | 13:01 |
babbageclunk | dimitern: thx! | 13:01 |
rick_h_ | natefinch: I just sent an email off to chuck to clarify and make sure I'm understanding things correctly | 13:01 |
natefinch | rick_h_: cool | 13:02 |
rick_h_ | natefinch: while that goes on how about looking at https://bugs.launchpad.net/juju/+bug/1620056 please | 13:02 |
mup | Bug #1620056: constraints should support cores=X <juju:Triaged by rharding> <https://launchpad.net/bugs/1620056> | 13:02 |
rick_h_ | natefinch: need to create a card on the board for it | 13:03 |
natefinch | rick_h_: ok, will do | 13:03 |
natefinch | heh, I was like "don't we already support that?" but it's a name change. I get it. | 13:03 |
rick_h_ | natefinch: ty | 13:03 |
rick_h_ | natefinch: right, little tweaking | 13:03 |
natefinch | rick_h_: the main problem is that the best interface for resource get would be to return two values: the path of the file on disk, and a boolean indicating whether or not it has been updated. But that's kind of hard to do on the command line. | 13:25 |
rick_h_ | natefinch: right, so that leads to two commands | 13:26 |
natefinch | rick_h_: but if you're running two commands already, you might as well just use md5sum | 13:27 |
rick_h_ | natefinch: but thing there is that folks can/will use different tools/etc. Making things like building a charm with layers and such more complicated. What mechanism did this guy use to hash/etc | 13:28 |
rick_h_ | natefinch: if it's a common use case thing we can make an opinionated standard that everyone just follows | 13:28 |
rick_h_ | natefinch: honestly, I want something like a return code from resource-get that says "did the file on disk change or not" | 13:28 |
rick_h_ | but I have a feeling anything non-0 might be more problematic for folks to rely on and build on | 13:29 |
babbageclunk | fwereade: LGTM | 13:29 |
natefinch | rick_h_: that's what I was thinking too.... if it's behind a new flag, it's probably not a big deal, since we wouldn't be breaking backwards compatibility. like resource-get --check-new | 13:30 |
natefinch | rick_h_: probably most people would just use whatever function we put in charm helpers | 13:30 |
natefinch | rick_h_: replied to your email. The other option is to have a --yaml flag (or --json) that outputs structured data, so we can return the path and a boolean | 13:35 |
rick_h_ | voidspace: fwereade ping for standup | 14:01 |
rick_h_ | dimitern: ping for standup | 14:01 |
dimitern | omw | 14:01 |
babbageclunk | hmm - anyone have advice about how to get into a 2nd gen x1 carbon to replace the ssd? | 14:01 |
voidspace | rick_h_: omw | 14:02 |
mgz | wat | 14:02 |
mgz | why you drop me google | 14:02 |
alexisb | babbageclunk, ping | 14:23 |
babbageclunk | alexisb: pong | 14:25 |
alexisb | heya babbageclunk, sorry moving slowish this morning | 14:25 |
alexisb | did you want to touch base this morning? | 14:26 |
babbageclunk | I forgot I was sitting in the hangout - camera's been on while I've been replacing my ssd. | 14:26 |
alexisb | lol | 14:26 |
alexisb | ok I will pop on | 14:26 |
* dimitern is back | 14:28 | |
voidspace | fwereade: was it you I'm due to be reviewing something for? | 14:58 |
fwereade | voidspace, http://reviews.vapour.ws/r/5637/ | 14:59 |
fwereade | voidspace, ta | 14:59 |
babbageclunk | voidspace: If you're looking for a pretty relaxing Friday afternoon-style review you can look at this one: http://reviews.vapour.ws/r/5638/ | 15:09 |
voidspace | babbageclunk: hah, because I have nothing else to do :-p | 15:09 |
voidspace | fwereade: looking | 15:09 |
babbageclunk | Ok, so do I want to do full-disk encryption? Will I still be able to install Windows for indie gaming fun? | 15:13 |
babbageclunk | Duh, no, that's what full-disk means. | 15:14 |
natefinch | babbageclunk: in my opinion, full disk excryption will just make your life harder. In theory, if Something Bad™ happens, you'll have protection against the government grabbing your laptop and using its contents against you. However, it doesn't prevent said government from hitting you with a pipe until you give them your password. | 15:14 |
dimitern | babbageclunk: LGTM, sorry it took so long | 15:19 |
babbageclunk | natefinch: true. | 15:19 |
babbageclunk | dimitern: No worries - sounds like there was some other stuff going on! | 15:20 |
babbageclunk | dimitern: Thanks! | 15:20 |
dimitern | babbageclunk: yeah, :/ a friday MAAS mystery (or likely even misery) | 15:21 |
dimitern | mgz: how goes? | 15:21 |
babbageclunk | dimitern: stink | 15:21 |
mgz | dimitern: doing some futzing still, will yell in a sex | 15:24 |
mgz | -x+c | 15:24 |
natefinch | wow, that really changes the meaning of your sentence | 15:25 |
dimitern | ;) | 15:25 |
mgz | ;_; | 15:25 |
dimitern | mgz: ok | 15:25 |
* rick_h_ goes for long lunch with family today, biab | 15:28 | |
mgz | dimitern: running now | 15:29 |
dimitern | mgz: I'm tailing the maas logs in the meantime | 15:29 |
mgz | dimitern: state server coming up on 10.0.30.13 | 15:34 |
mgz | you should be able to ssh in there with the ci staging key shortly | 15:34 |
dimitern | mgz: yeah, trying now | 15:34 |
mgz | dimitern: okay, we're at the fail point | 15:46 |
dimitern | mgz: no errors so far in the controller log | 15:47 |
voidspace | fwereade: LGTM | 15:56 |
dimitern | mgz: ok, got the error - looking earlier in the logs | 15:56 |
fwereade | voidspace, ta | 15:56 |
dimitern | mgz: /var/log/postgresql/postgresql-9.3-main.log on maas has some very interesting errors | 16:10 |
dimitern | frobware: ^^ | 16:10 |
redir | morning juju | 16:13 |
frobware | dimitern: categorically works for me on 1.9 | 16:14 |
dimitern | frobware: it works for me on both 1.9 and 2.0 | 16:14 |
frobware | dimitern: hmm; I have a few teething problems on my maas 2.0 setup atm | 16:22 |
dimitern | frobware, mgz: I think I nailed it | 16:26 |
dimitern | the db got corrupted | 16:26 |
dimitern | when the upgrade was done yesterday | 16:27 |
mgz | dimitern: what's up with having both 9.5 logs (last from -09-08) and 9.3 logs (current)? | 16:27 |
dimitern | instead of upgrading from 2.0.0rc3 to 2.0.0, it was done first to 2.1.0alpha3, then downgraded to 2.0.0 final | 16:27 |
mgz | dimitern: aha | 16:28 |
dimitern | unfortunately the alpha3 package ran some db migrations, which dropped maasserver_node_system_id_seq | 16:28 |
mgz | the migrations *should* be two ways | 16:28 |
dimitern | and that's causing the device creation to fail | 16:28 |
dimitern | I bet I can repro it manually even with the maas cli only | 16:29 |
dimitern | pgsql 9.3 is the one maas uses, the newer was installed with the intent of upgrading the maas one, but that didn't happen | 16:29 |
dimitern | all those hints can be observed in /var/log - ./apt/term|history.log, ./postgresql/.. but unfortunately not in any the maas logs | 16:31 |
dimitern | frobware, mgz: and indeed I did - http://paste.ubuntu.com/23155215/ | 16:34 |
dimitern | that maasserver_node_system_id_seq got dropped months ago after 2.0 switched to shorter node names (check the MP - the only google result for the seq name) | 16:36 |
dimitern | I guess nobody seriously thought about the existing maas 1.9 node ids will be a problem after upgrading to 2.0 and then to 2.1 | 16:37 |
frobware | dimitern, mgz: so I just don't run into this because I don't upgrade... :( | 16:42 |
frobware | dimitern, mgz: bitter experience from Windows 95. | 16:42 |
dimitern | frobware: well isn't it better not to run into these issues? :) | 16:45 |
frobware | dimitern: yes, but customers (you would hope) are using our older versions... | 16:46 |
dimitern | mgz: I'm trying to see if it's salvageable without reinstall | 16:46 |
dimitern | by reverting the 2.1.0 migrations | 16:46 |
mgz | dimitern: thanks... | 16:46 |
frobware | dimitern, mgz: funnily enough I am having psql problems installing a fresh version of maas 2.0 | 16:47 |
frobware | dimitern: that's my third install which has failed. | 16:49 |
rick_h_ | frobware: dimitern any progress on identifying the issue? | 17:29 |
frobware | rick_h_: see issues above about psql failures in migrations | 17:30 |
dimitern | rick_h_: yeah | 17:30 |
rick_h_ | frobware: reading backlog, so there's issues with upgrading and how the ids change over time? | 17:30 |
rick_h_ | frobware: dimitern k, sounds like we need to file a bug against MAAS? | 17:30 |
rick_h_ | frobware: dimitern and then to update our install to be a fresh install of MAAS in CI and we should pass tests cleanly? | 17:31 |
rick_h_ | mgz: ^ ? | 17:31 |
dimitern | rick_h_: botched maas upgrade 2.0r3->2.1.0a1->2.0.0 corrupted the dn | 17:31 |
dimitern | db | 17:31 |
rick_h_ | dimitern: oh, it was a botched upgrade? | 17:31 |
rick_h_ | dimitern: ok | 17:31 |
dimitern | some db migrations ran, but not all are reversible | 17:31 |
rick_h_ | sinzui: mgz so are we able to clean the MAAS and rerun? | 17:31 |
frobware | rick_h_: agreed to the clean install; leave the other as-is for continued investigation | 17:31 |
frobware | mgz: ^^ | 17:31 |
dimitern | rick_h_: I was trying to manually revert those 2.1.0 migrations, but not all can be reverted | 17:32 |
rick_h_ | dimitern: ok, let's not bother with that imo | 17:32 |
dimitern | ok | 17:32 |
sinzui | dimitern: rick_h_ ouch, and the error happened before the botched upgrade | 17:32 |
rick_h_ | sinzui: right, but it was a different error | 17:32 |
rick_h_ | sinzui: that was backed out with the revert, but then we kept failing with a new error due to the botched upgrade | 17:32 |
dimitern | sinzui: yeah, a bit more complicated than usual | 17:32 |
rick_h_ | sinzui: mgz dimitern frobware hangout to set the path forward? | 17:33 |
sinzui | rick_h_: dimitern: I don't know what clean up means in this case. A clean install is something I have never done. I wont promise ot for today | 17:33 |
dimitern | rick_h_: sure - standup HO? | 17:33 |
rick_h_ | dimitern: rgr | 17:33 |
dimitern | I need to go soon | 17:33 |
mgz | I think we need a new maas vm | 17:33 |
frobware | sinzui: I can help with a clean install | 17:33 |
mgz | or some more serious help from a maas expert | 17:33 |
frobware | mgz: agreed | 17:33 |
rick_h_ | alexisb: BrettD mgz sinzui welcome to join https://hangouts.google.com/hangouts/_/canonical.com/core?authuser=1 | 17:33 |
alexisb | rick_h_, not atm | 17:34 |
rick_h_ | alexisb: k, will fill you in afterwards then | 17:34 |
sinzui | mgz hop back up | 17:46 |
sinzui | back on | 17:46 |
sinzui | mgz: really, hop back on to https://hangouts.google.com/hangouts/_/canonical.com/core?authuser=1 | 17:48 |
mgz | sinzui: omw | 17:48 |
marcoceppi | natefinch: what version of juju did you disable ciphers | 18:01 |
marcoceppi | like the shitty insecure ciphers | 18:01 |
rick_h_ | marcoceppi: https://bugs.launchpad.net/juju/+bug/1604474 b16 | 18:01 |
mup | Bug #1604474: Juju 2.0-beta12 userdata execution fails on Windows <azure-provider> <ci> <juju2.0> <oil> <oil-2.0> <regression> <vpil> <windows> <juju:Fix Released by natefinch> <https://launchpad.net/bugs/1604474> | 18:01 |
natefinch | marcoceppi: uh | 18:02 |
marcoceppi | rick_h_: nice, thanks | 18:03 |
natefinch | yes that | 18:03 |
sinzui | rick_h_: mgz: functional-container-networking does pass on maas 1.9. maas 2.0 is being rebuilt now | 18:15 |
rick_h_ | sinzui: ty | 18:16 |
rick_h_ | sinzui: feel a bit better about things now | 18:16 |
natefinch | method returns value, error. The only place we call the method... we ignore the error. FanTASTIC. | 18:22 |
lazyPower | silly question, i've got an application depoyed, its a subordinate, and its active in the "unit view", but its not listed as active in the "app view" - is this known an expected? http://imgur.com/AGp83ik | 20:27 |
natefinch | lazyPower: dunno... there's been a lot of churn in that area. Not sure who last updated that code. | 20:28 |
lazyPower | ok, i figured it was intentional as subordinates dont technically occupy a unit, they co-locate.. but it was a bit startling of a realization | 20:29 |
lazyPower | thanks natefinch | 20:29 |
natefinch | lazyPower: I know we intentionally hide them in some parts... not sure if this was an intentional part or not :) | 20:29 |
redir | ready for a review if anyone is still around: http://reviews.vapour.ws/r/5640/ | 22:01 |
redir | bbiab | 22:02 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!