babbageclunk | wallyworld: were there any conflicts backporting it? | 00:36 |
---|---|---|
wallyworld | babbageclunk: yup, due to 2.8 supporting cancelling actions | 00:39 |
wallyworld | babbageclunk: were you still going to review the PR? | 01:44 |
thumper | wallyworld: does this bug look familiar? https://bugs.launchpad.net/juju/+bug/1828076 | 02:42 |
mup | Bug #1828076: juju deploy fails because no machines can be found in MaaS in a given AZ <canonical-bootstack> <juju:Triaged> <https://launchpad.net/bugs/1828076> | 02:42 |
wallyworld | thumper: not really. there was an AZ issue in AWS that was fixed, but I am not aware of any recent maas fixes | 02:44 |
wallyworld | kelvinliu: tlm: got time for a HO? | 02:45 |
kelvinliu | wallyworld: yep | 02:46 |
tlm | yeah | 02:46 |
thumper | wallyworld: was it a general az fix or aws specific? | 02:47 |
wallyworld | thumper: i thought aws but can check | 02:47 |
wallyworld | thumper: i was wrong, it was generic code | 03:00 |
wallyworld | so the bug could well be fixed | 03:01 |
wallyworld | ws tested on aws as that's where the issue was reported | 03:01 |
kelvinliu | wallyworld: the hook failed because the leader was gone. the leader was gone because the map of pod <-> unit was lost. | 03:13 |
wallyworld | kelvinliu: otp, sorry | 03:13 |
kelvinliu | nw | 03:15 |
wallyworld | kelvinliu: yeah, so seems like a real issue to fix | 03:35 |
wallyworld | kelvinliu: since xtian is away, could you please review this backport whenever convenient? https://github.com/juju/juju/pull/11430 | 03:35 |
kelvinliu | but I wasn't able to re-produce the pvc terminating error | 03:36 |
kelvinliu | looking now | 03:36 |
tlm | i am wondering if the pvc thing came from a storage increase ? | 03:37 |
babbageclunk | wallyworld: oops got halfway through before I had to go - doing it now! | 03:38 |
wallyworld | tlm: you are not allow to change storage config with juju once deployed. not to say there isn't a bug there somewhere | 03:38 |
babbageclunk | ha, it just updated that kelvinliu has, sorry | 03:38 |
wallyworld | babbageclunk: all good, i thought you were away for the arvo! | 03:38 |
tlm | ok just not sure where that 1g to 20g came from | 03:39 |
babbageclunk | no, back | 03:39 |
tlm | may have read the logs wrong | 03:39 |
wallyworld | tlm: we use 20G for the controller PV | 03:39 |
tlm | ah k | 03:39 |
wallyworld | 1G by default for workload | 03:39 |
kelvinliu | done | 03:39 |
wallyworld | tyvm | 03:39 |
kelvinliu | np | 03:40 |
wallyworld | i will tets on iaas to be sure we didn't break anything there | 03:40 |
wallyworld | kelvinliu: what file did the golang version selection for building debs live in? | 04:06 |
kelvinliu | wallyworld: https://bazaar.launchpad.net/~juju-qa/juju-release-tools/packaging-juju2-default/view/head:/debian/control#L8 | 04:09 |
wallyworld | ty | 04:10 |
kelvinliu | np | 04:10 |
stickupkid | manadart, ping | 08:13 |
manadart | stickupkid: Pongzor. | 08:15 |
stickupkid | daily | 08:15 |
elox70 | Watching from Sweden! | 09:07 |
elox70 | I'have a question to the live show. Can I ask it here ? | 09:13 |
stickupkid | elox70, sure | 09:15 |
elox70 | Cool! I'm also on youtube. | 09:17 |
elox70 | Thanx for the great show! | 09:48 |
manadart | stickupkid: Simple one: https://github.com/juju/juju/pull/11432 | 10:42 |
stickupkid | manadart, we've already got an IDSet | 10:43 |
stickupkid | manadart, probably needs renaming https://github.com/juju/juju/blob/develop/core/network/subnet.go#L171 | 10:43 |
manadart | stickupkid: That's what I have renamed. | 10:44 |
stickupkid | ah | 10:44 |
* stickupkid is an idiot | 10:44 | |
stickupkid | manadart, that's actually our dependency list of reload spaces https://paste.ubuntu.com/p/5PWFKnqjFp/ | 11:30 |
manadart | stickupkid: Okay. | 11:32 |
manadart | stickupkid: Another precursor patch for my validation work: https://github.com/juju/juju/pull/11433 | 12:34 |
hml | achilleasa: got the last resolver test done - it wasn’t testing what it thought it was testing. now it does. :-). i’ll push the state manager change shortly | 14:00 |
manadart | achilleasa: Want to do QA on 11431? Code looks good, but I am running out of disk here and reset my mk8s. | 14:12 |
hml | achilleasa: pushed, working on the state changes now | 14:13 |
achilleasa | manadart: ok, I will try it locally in about 5 min | 14:20 |
rick_h_ | manadart: stickupkid please let me know when the reviews/merge is ready for wallyworld's patch please. | 14:37 |
achilleasa | rick_h_: running QA steps for ^^ | 14:37 |
rick_h_ | cool ty achilleasa | 14:38 |
stickupkid | manadart, I've really picked one thread here, there is some really funky stuff here. | 14:53 |
manadart | stickupkid: That's the Juju way. | 14:55 |
stickupkid | essentially we "shim" state, but actually it does different things, it's like an adapter | 14:55 |
stickupkid | I might rename it to adapter and this it won't lie | 14:55 |
hml | achilleasa: I think i have all of the changes made for the relation pr | 15:04 |
hml | achilleasa: pondering the drive by to remove the panic. wondering side effects: what might be left behind because of an error rather than a restart of the uniter caused by the panic. | 15:05 |
achilleasa | hml: I think if you bubble up the error the uniter will restart anyway. However, a panic (unless the dependency engine catches and suppresses panics) would likely cause the agent to be restarted, right? | 15:07 |
hml | achilleasa: i’d like to be sure. would rather not do in a drive by | 15:07 |
achilleasa | it's probably fine to leave as-is for now | 15:07 |
hml | achilleasa: will look at in a separate pr | 15:08 |
achilleasa | however, we should probably get rid of them when we see them again in the future | 15:08 |
achilleasa | panics and long-running tasks are probably a bad combination ;-) | 15:08 |
achilleasa | rick_h_: QA for 11431 complete and PR has 2 ticks; we can prob go ahead and land it | 15:25 |
rick_h_ | achilleasa: awesome ty | 15:28 |
achilleasa | hml: reviewing 11414 | 15:28 |
hml | achilleasa: i have a few tests in state to clean up | 15:52 |
achilleasa | hml: got some questions about error handling; overall it looks great so far. Do you want me to get started on the QA steps? | 15:53 |
achilleasa | otherwise I will go back to my caas tests | 15:53 |
hml | achilleasa: yes please on the QA steps, i’ve added some more detail there | 15:54 |
achilleasa | hml: ok! | 15:55 |
hml | achilleasa: reading questions | 15:55 |
achilleasa | hml: QA works as expected. Nice! | 16:12 |
hml | achilleasa: sweet | 16:12 |
hml | achilleasa: updating the error handling | 16:12 |
achilleasa | also, I think I figured out why my k8s upgrade was not picking up the fix. I will verify now... | 16:13 |
hml | achilleasa: i have to clear any operator things from the local docker sometimes… | 16:14 |
achilleasa | I think it does not use AlwaysPull so it never pulled the new image from dockerhub | 16:14 |
hml | achilleasa: i’m thinking the qa should deal with some subordinates also, that has different code paths | 16:20 |
hml | achilleasa: pushed up all my changes… including some error stuff. i have a gremlin in mockSteps28Suite.TestMoveUnitAgentStateToControllerIAAS setup code that i have to nail, but wanted to get the rest up for yo | 16:58 |
achilleasa | hml: looking | 16:59 |
achilleasa | hml: error handling looks correct to me. Given that it's just a test that needs fixing do you want me to approve the PR now before my EOD? | 17:03 |
hml | achilleasa: yes please! and thank you for the indepth review on a large pr | 17:04 |
hml | playing whack a mole with setup directory creation. | 17:04 |
achilleasa | hml: done. After merging please run your QA steps again in conjunction with the ones from 11423 to make sure that both upgrades and model migrations work without issues | 17:07 |
hml | achilleasa: rgr | 17:07 |
hml | achilleasa: were are we with the juju upgrades for caas investigation? | 17:08 |
wallyworld | achilleasa: docs updated for k8s upgrade testing | 21:42 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!