/srv/irclogs.ubuntu.com/2019/11/08/#juju.txt

thumperhere's a very boring review for someone: https://github.com/juju/juju/pull/1088403:45
thumperlong and boring and almost all mechanical03:45
thumper157 files, +764 −62203:46
* thumper packs up for sprint03:46
anastasiamacwallyworld: thank you for the review! r u k with my take o re-phrase?03:50
wallyworldlooking03:54
wallyworldanastasiamac: made a suggestion03:58
anastasiamacwallyworld: i know that historically we have only updated regions... however potentially it can b other properties too... like auth-types...04:04
anastasiamacwallyworld: i think i'd rather just say 'to update public clouds ...' without any further detail04:04
wallyworldyeah true. not too fussed on that bit of the wording, although region is really what it's about 99% of the time. more fussed with "from a central location" bit. doesn't really matter how, users just need to know it will happen04:05
anastasiamacta04:05
wallyworldhpidcock: kelvinliu: bug 1851763 may be a rc stopper. it may just be CK on maas. microk8s worked for me this morning and i think CK on say AWS or Azure works. could one of you please confirm that is the case so we can narrow this down to just a maas issue. i need to get stuff done for the sprint04:54
mupBug #1851763: [2.7 rc2] can't deploy k8s-based model re-using cloud controller <juju:New> <https://launchpad.net/bugs/1851763>04:54
kelvinliuwallyworld: im looking it now04:56
wallyworldty, i need to get stuff sorted if i can get away from the keyboard05:01
kelvinliunp,05:05
kelvinliuwallyworld: I tested, lxd controller <- microk8s; IaaS controller <- azure CK; IaaS controller <- aws CK; all working, so I think it should be just maas and not related with CaaS.06:25
wallyworldkelvinliu: gr8 tyvm. the suspicion is that we are copying across the spaces for the controller cloud when a new model is added, and not doing a check that the model uses the same cloud06:29
wallyworldmaas controller cloud has spaces, k8s cloud doesn't06:29
kelvinliubriefly look on apiserver/facades/client/modelmanager/modelmanager.go, didn't find any space related change.06:37
wallyworldjoe will look into it06:41
kelvinliuthanks06:44
gnuoyHi, we just switched over to using juju 2.7 candidate and out charm testing broke. The problem is that when test a charm we use a bundle which uses all supporting  charms from the charm store and uses the local copy of the charm undertest. We reference the charm undertest in the bundle like this:09:38
gnuoycharm: ../../../nova-cell-controller09:38
gnuoybecause the overlay with that in lives inside the charm here:09:39
gnuoynova-cell-controller/tests/bundles/overlays09:39
gnuoyand the local paths seem  to relative to the overlay location09:40
gnuoythis worked in 2.6 but in sit I get:09:40
gnuoyERROR cannot deploy bundle: the provided bundle has the following errors:09:40
gnuoycharm path in application "nova-cell-controller-cell2" does not exist: /nova-cell-controller09:40
gnuoySoes anyone know of a 2.7 change that might account for this ?09:40
gnuoys/Soes/Does/09:40
stickupkidgnuoy, nope, but a bug will be useful09:43
stickupkidgnuoy, https://bugs.launchpad.net/juju/+bugs?field.tag=bitesize09:44
stickupkidsorry ignore the bitesize query part09:44
stickupkidhttps://bugs.launchpad.net/juju/+filebug09:44
gnuoystickupkid, thanks. tbh  if a full qualified path works in 2.6 I may just switch to doing that.09:44
gnuoy(seems to work  in 2.7)09:44
nammn_demanadart rick_h stickupkid Pr is now in a reviewable state. Happy to get some input! https://github.com/juju/juju/pull/1086709:54
stickupkidtraining day09:54
nammn_destickupkid: ahhh you right its friday09:59
manadartNeed a review for the k8s-on-MAAS fix: https://github.com/juju/juju/pull/1088612:59
achilleasamanadart: looking13:00
achilleasamanadart: why is AllSpaces returning no subnets on maas even though they seem to have space IDs in the DB?13:01
manadartachilleasa: https://github.com/juju/juju/blob/bdaefb0a07d6b033ef5c3499085889fb9f8fccb1/state/spaces.go#L98 :)13:04
achilleasashould I fix it now?13:05
manadartachilleasa: Yep.13:05
achilleasamanadart: hmmm... space.Subnets() may potentially return an error...got to change the return signature for NetworkSpace() :-(13:15
manadartSimple one: https://github.com/juju/juju/pull/1088714:29
rick_hmanadart:  is the space removal bits intended?14:31
manadartrick_h: Yeah, that was an indirection for space lookups that no longer happen for client API addresses.14:32
rick_hall good, I first read that as a "fix typo" and then saw deleted code but see the full comment on there14:32
rick_h+114:32
achilleasamanadart: so, now the machiner will only update the linklayer devices while the instancepoller will update the provider addresses, correct?14:34
manadartachilleasa: The instancepoller will need to merge link-layer data if it differs from what is there. Otherwise we won't have provider IDs for any of that data.14:36
manadartIn addition to setting those machine collection addresses.14:36
achilleasamanadart: gotcha14:38
danboidIs there a dedicated irc channel for Ubuntu openstack?14:42
danboidI've been unable to SSH into any instances I have deployed with Ubuntu openstack14:43
rick_hdanboid:  definitely https://help.ubuntu.com/community/OpenStack14:45
danboidThe web UI says it has deployed successfully but looking at the log it doesn't look like the vm is being assigned an IPv4 address (contrary to what the web UI says) nor is it copying my SSH key14:45
rick_hoh hmm, guess that's just openstack. I know they've also got something else14:45
rick_hthere's this but that's more for the charming/operations side but might be able to help https://docs.openstack.org/charm-guide/latest/find-us.html14:46
danboidrick_h, Thanks, I'll try there14:50
nammn_derick_h:  here is the pr where i need some input for qa/ux related things https://github.com/juju/juju/pull/1086715:01
rick_hty nammn_de15:06
nammn_derick_h: I tried to follow the doc spec as much as possible :D15:06
rick_hsweet15:08
rick_hbuilding now15:08
aisraelrick_h: Do you know when 2.7 rc2 was released?15:42
aisraelnm, looks like yesterday15:43
rick_haisrael:  yesterday?15:43
* rick_h can't keep track of time15:43
rick_haisrael:  one know issue causing an rc3 monday morning for a corner case of k8s models on top of maas contorller https://bugs.launchpad.net/juju/+bug/185176315:44
mupBug #1851763: [2.7 rc2] can't deploy k8s-based model re-using cloud controller <juju:Fix Committed by manadart> <https://launchpad.net/bugs/1851763>15:44
aisraelrick_h: okay. I saw there's an rc3 candidate in edge. We're chasing a couple potential bugs right now and will file them asap15:45
nammn_destickupkid: the acceptancetest fix regarding cli https://github.com/juju/juju/pull/1088815:45
aisraelAll related to k8s15:45
nammn_desmall one15:45
rick_haisrael:  sounds good thanks for filing15:45
rick_haisrael:  ty for testing the RC, much better to find stuff there.15:46
rick_hnammn_de:  not sure it's direct in your PR but heads up https://bugs.launchpad.net/juju/+bug/185185316:00
mupBug #1851853: a rolled back commit shows as a commit in list-commit <juju:Triaged> <https://launchpad.net/bugs/1851853>16:00
rick_hnammn_de:  doing a proper test with a different charm that has config this time16:00
rick_hnammn_de:  but I think that commit 0 might be a real issue16:01
nammn_derick_h: thanks! Oh ahhh, I think I have a small error in the query, let me change that. I suspect that i should have not searched for completed16:12
nammn_delet me fix that16:12
rick_hnammn_de:  cool16:12
rick_hI'll keep poking16:13
rick_hmmmm, back to non-dark roast coffee so yummy :P16:13
nammn_deit should work for not aborted ones, because as it now, i think, it will always show aborted ones with id ==0 :D16:13
rick_hnammn_de:  oh lol16:13
manadartnammn_de: bson.M{"completed": bson.M{"$gte": 1}} should be generation-id instead of completed.16:16
nammn_derick_h manadart: yeah thought so as well,  patch is up :D16:19
nammn_demanadart: thansk for taking the quick look!16:19
rick_hnammn_de:  ok, rebuilding/back to poking ty16:20
rick_hnammn_de:  ok, ty this feels really cool to play with16:54
rick_hnammn_de:  feedback inbound, and the rolled back ones are good now with the patch16:54
* rick_h is trying to think if we have anything else in juju that's date ordered16:55
rick_hI like the commits in reverse order but don't want to flip/flop if we do it another way somewhere else...but I can't think of where we would16:55
nammn_derick_h: thanks for the feedback will work it in next week.  list-users cuts off the output after the day. I can change that to that as well17:04
achilleasarick_h: perhaps a --sort=asc/desc flag?17:04
nammn_deachilleasa: sounds good to me, I can add that17:04
rick_hnammn_de:  oh yea, definitely don't want to just keep a day17:05
achilleasanammn_de: I am pretty sure git has a similar flag for logs17:05
rick_hnammn_de:  yea, let's not do the flag yet17:05
rick_hnammn_de:  but just desc and we'll tweak it after using it more for longer periods of time17:05
nammn_dehow "much" of the date do you think makes the most sense for list-commit?17:06
nammn_de2019-11-06  || 2019-11-06 hh:ss ?17:06
nammn_de*hh:mm17:07
rick_hnammn_de:  so I assumed the human readable thing did 12m ago, 4hrs ago, 10 days ago?17:07
nammn_derick_h: ahhh sure17:07
rick_his that not what it's doing? when I looked on my demo controller it was saying "12m ago" and figured it was human readable-ling17:07
nammn_deits doing both,  if its further away than a timeline it shows the yyyy-mm-dd, else the human readable thingz17:08
nammn_dethats why i got confused17:08
rick_hnammn_de:  ah, ok that's find17:08
rick_herr fine17:09
rick_hnammn_de:  so cheat off what that's doing please :)17:09
nammn_derick_h: will do, cheating off is my skill17:09
manadartrick_h: Backport service removal script: https://github.com/juju/juju/pull/1088917:50

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!