tlm | wallyworld: got 5 minutes for HO problem ? | 03:41 |
---|---|---|
wallyworld | sure soon | 03:42 |
wallyworld | tlm: free now | 04:00 |
tlm | k | 04:00 |
babbageclunk | wallyworld: https://github.com/juju/juju/pull/11543 | 06:08 |
babbageclunk | I think it might have a problem with something in feature tests that rely on legacy leases though. | 06:08 |
wallyworld | babbageclunk: looking | 06:38 |
wallyworld | babbageclunk: yeha, there's a MachineLegacyLeasesSuite in cmd/jujud/agent machine_test | 06:40 |
wallyworld | so a few tests to fix | 06:40 |
hpidcock | wallyworld: quick pr https://github.com/juju/juju-db-snap/pull/2 | 07:30 |
wallyworld | sure | 07:30 |
wallyworld | lgtm | 07:31 |
hpidcock | danke | 07:31 |
flxfoo | Hi guys, sorry to interrupt again. But someone is asking me some questions like | 07:59 |
flxfoo | Is JUJU able to provision AWS resources or is managing only application deployment/layer? | 08:00 |
flxfoo | and | 08:00 |
flxfoo | If able to deploy AWS resources, is it able to deploy autoscaling groups? Or just fixed defined number of instances? | 08:00 |
flxfoo | if anyone has some answers... you are welcome :P | 08:00 |
timClicks | fixfoo: provision, yes to autoscaling | 08:01 |
timClicks | except it's not autoscaling as per kubernetes, it's manual | 08:01 |
timClicks | e.g. you ask juju to change the application's scale, and it will update all config on your behalf | 08:01 |
flxfoo | timClicks: thanks! | 08:05 |
flxfoo | still don't know what they are calling "groups" | 08:05 |
=== salmankhan1 is now known as salmankhan | ||
jam | timClicks[m], are you around? | 10:35 |
manadart | achilleasa: Got a couple of small patches to come in under the one I am working. | 11:10 |
manadart | This is the first: https://github.com/juju/juju/pull/11545 | 11:11 |
manadart | achilleasa: And this is the other: https://github.com/juju/juju/pull/11546 | 11:15 |
achilleasa | manadart: looking | 11:16 |
hpidcock | PR for someone https://github.com/juju/testing/pull/148 | 11:34 |
hpidcock | another PR https://github.com/juju/juju/pull/11547 | 11:44 |
achilleasa | hml: got any time to look at 11547 ^^ | 12:14 |
hml | achilleasa: not really, i was thinking we needed to look at it though from email. | 12:14 |
hml | achilleasa: i’m tight on time - read thumper’s email on RC1 | 12:14 |
hml | :-D | 12:14 |
achilleasa | hml: ah... that email! No worries, I will take a look then | 12:15 |
hml | achilleasa: ty | 12:15 |
hml | achilleasa: we should look at 11544 also | 12:27 |
pmatulis | if i reboot a juju machine and immediately send it a command like 'juju upgrade-series 0 complete' is that guaranteed to work or do i need to wait? | 13:30 |
pmatulis | hml, any idea? ^^^ | 14:06 |
hml | pmatulis: you could have a race, if the machine doesn’t go down fast enough, i’d think. cannot guarantee. | 14:08 |
hml | pmatulis: a bit swamped at the momment to dig into how our CI tests this | 14:08 |
hml | manadart: do you have an easy answer for pmatulis ^^^ ? | 14:09 |
manadart | pmatulis hml: You should be able to send the command immediately. It needs the agent to do the work, so the command will implicitly wait for the worker to do its thing. | 14:11 |
manadart | This is actually a known issue with it - reliance on a well connected client should not be a thing. | 14:11 |
pmatulis | manadart, you're saying it will work out fine as long as the client maintains connectivity with the controller? | 14:13 |
manadart | pmatulis: Yes, there is an open bug around it. | 14:14 |
manadart | https://bugs.launchpad.net/juju/+bug/1855013 | 14:14 |
mup | Bug #1855013: upgrade-series hangs, leaves lock behind <seg> <juju:Triaged> <https://launchpad.net/bugs/1855013> | 14:14 |
pmatulis | manadart, thanks a lot. i don't see anything about bad client:controller connectivity as the cause though | 14:17 |
manadart | pmatulis: The symptoms line up with what we know about how the feature was written. | 14:19 |
pmatulis | manadart, alright | 14:21 |
manadart | achilleasa, hml, petevg: This is finally ready to go now: https://github.com/juju/juju/pull/11534 | 14:41 |
petevg | Woot! | 14:45 |
petevg | achilleasa, manadart: if either of you have time to take a look at hml's draft for https://github.com/juju/juju/pull/11542 before your eod today, I'd appreciate it. That would help us try to sneak it in to the rc1. | 14:46 |
manadart | petevg: I can look now. | 14:47 |
petevg | +1. Thx! | 14:47 |
achilleasa | petevg: I ... think ... I have finally managed to reproduce the kvm bug | 14:54 |
petevg | achilleasa: nice! Was it a caching issue, or something else? | 14:54 |
achilleasa | petevg: still going through my logs but I believe after we get the metadata from the stream and before we fetch the actual image, the base URL (which should point to the model-config) is left empty... | 14:56 |
petevg | Interesting. | 14:56 |
achilleasa | petevg: so if it's not correctly propagated, we get here: https://github.com/juju/juju/blob/develop/environs/imagedownloads/simplestreams.go#L77-L79 | 15:03 |
achilleasa | and this is why they fall back to cloud-images.ubuntu.com | 15:04 |
achilleasa | looks like the streams index was fetched from the correct location but the image used the fallback | 15:04 |
petevg | Aha. | 15:04 |
hml | manadart: ty for 11542, I pushed up some commands testing. there is an acknowledged short cut….. | 15:42 |
hml | achilleasa: manadart how do i get the skipped github actions to run after converting a pr from draft to open? | 15:46 |
hml | haven’t done this since we changed it | 15:46 |
hml | no buttons are appearing to me. | 15:47 |
pmatulis | manadart, hi, i got a similar question re actions. is there a timing concern for a unit that has had an action applied if the --wait option is omitted? | 16:15 |
pmatulis | (my use case is pausing a unit prior to 'upgrade-series prepare' for the associated machine) | 16:18 |
hml | pmatulis: what command are you using with —wait? | 16:19 |
hml | pmatulis: show-action-output | 16:19 |
hml | ? | 16:19 |
pmatulis | hml, run-action | 16:19 |
pmatulis | (to pause the unit) | 16:20 |
hml | pmatulis: i’d recommend checking the action status at the very least. | 16:20 |
hml | there is also a show-action-status | 16:20 |
pmatulis | hml, realise that. just wondering if there is indeed a timing issue | 16:20 |
pmatulis | (potentially) | 16:21 |
pmatulis | to be clear, if keystone/2 is on machine 0 can i perform these two commands safely without waiting? | 16:24 |
pmatulis | juju run-action keystone/2 pause | 16:24 |
pmatulis | juju upgrade-series 0 prepare bionic | 16:24 |
achilleasa | petevg: are there any plans for another 2.6 release? I have a fix for the kvm bug currently targeting 2.7 but can also do 2.6 -> 2.7 -> develop | 16:49 |
achilleasa | hml: approved but without QA as I am near EOD | 16:52 |
hml | achilleasa: ty, i’ll get petevg to do qa. :-) | 16:52 |
hml | achilleasa: do you know how to switch the github actions back on for a pr? | 16:52 |
achilleasa | hml: no :-( do you need admin permissions? | 16:53 |
hml | achilleasa: dunno? i’ll check with hpidcock later | 16:53 |
petevg | achilleasa: I don't know whether we're planning another 2.6 release or not. Let's just target 2.7 and develop with the fix for now. | 16:54 |
achilleasa | hml: can you take a look at https://github.com/juju/juju/pull/11550? Perhaps petevg can lend a hand with the QA steps as well. Since I am EOW it would be great if you (or anyone from APAC) could forward-port (should be a clean merge) to develop and update the bug with a link to the other RP | 17:50 |
hml | achilleasa: looking | 17:50 |
hml | achilleasa: you used guimaas to test? | 17:50 |
achilleasa | hml: yes but the image server must be accessible by it (I used a VPS for testing) | 17:53 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!