| wallyworld | babbageclunk: is it true we hard code leases auto expire to false? did we have plans to change that? if auto expire is always false, then we can remove more code? | 00:10 |
|---|---|---|
| babbageclunk | wallyworld: ? not sure what you mean | 00:10 |
| wallyworld | / Autoexpire is part of the lease.Store interface. | 00:11 |
| wallyworld | func (*store) Autoexpire() bool { return false } | 00:11 |
| wallyworld | controls how the worker expires leases | 00:11 |
| wallyworld | whether it does it on the next tick or if a client call i needed | 00:11 |
| babbageclunk | Ah - right! Yes, I'll remove that - it's because the db leases needed to be expired, but the raft ones get expired on clock tick | 00:12 |
| babbageclunk | thanks for the reminder! | 00:12 |
| wallyworld | babbageclunk: np, but for raft, if it is done automatically, wouldn't auto expire be true? | 00:12 |
| babbageclunk | it is | 00:15 |
| babbageclunk | I mean, I'll remove the code in lease manager that does stuff when !autoexpire | 00:15 |
| wallyworld | but it's hard coded false? | 00:15 |
| babbageclunk | in the non-raft store | 00:16 |
| wallyworld | ah, ok, i was looking at the wrong one | 00:16 |
| wallyworld | duh | 00:16 |
| babbageclunk | yup yup | 00:16 |
| wallyworld | babbageclunk: you can also remove ExpireLease() right? and reduce code in provider dummyLeaseStore, or even remove that entirely | 00:46 |
| babbageclunk | yeah, that was what I was planning | 00:47 |
| thumper | https://github.com/juju/juju/pull/11582 for someone | 01:04 |
| thumper | backporting develop fix to 2.8 | 01:05 |
| thumper | should really have landed in 2.7, but I don't think it is worthwhile going that far back now | 01:05 |
| wallyworld | lgtm | 01:19 |
| wallyworld | thumper: wanna land in 2.8-rc though | 01:19 |
| wallyworld | that will get forward ported today later | 01:20 |
| thumper | wallyworld: ok | 01:20 |
| kelvinliu | wallyworld: hpidcock could any of you take a look this PR https://github.com/juju/juju/pull/11583 for fixing upgrade-charm? ty | 01:42 |
| wallyworld | sure | 01:43 |
| thumper | wallyworld, hpidcock, kelvinliu, tlm: do any of you have any running k8s models that have been upgraded from earlier versions? | 01:48 |
| * thumper wants to test a theory | 01:48 | |
| wallyworld | not currently | 01:55 |
| wallyworld | kelvinliu: +1 but with a request to fix the RemoteInitFunc signature to remove runningStatus. the cahce cleanup can happen later, but will be nioce to have the func clean | 02:04 |
| kelvinliu | wallyworld: we still need get the pod name from the status, or we can change the status from podName string. or do u think we can change it later when we decide the remove the status entirely? | 02:08 |
| kelvinliu | typo, change the status to podName string | 02:09 |
| tlm | thumper: I don't | 02:09 |
| wallyworld | kelvinliu: ah right, i missed that, sure | 02:09 |
| tlm | wallyworld: cooking with gas when you're ready | 02:09 |
| wallyworld | tlm: looking | 02:10 |
| kelvinliu | wallyworld: I think we just need to test for upgrading different workload type - stateless, stateful, daemon, but probably no need to test different k8s cloud, agree? | 02:12 |
| wallyworld | we should at some point but can be microk8s for now i think | 02:13 |
| wallyworld | tlm: SetPasswords() also needs changing to be an api that takes just the password. it can internally create the params.EntityPassword since we use a pluggin at the back end, but the api called should just pass in a single password | 02:15 |
| wallyworld | and result.OneError() | 02:15 |
| wallyworld | func sig just returns a single error, not ErrorResults | 02:15 |
| wallyworld | tlm: also, i added tag 2.0.1 to juju/description | 02:17 |
| wallyworld | i left comments in the pr | 02:22 |
| tlm | wallyworld: cheers doing now | 02:23 |
| babbageclunk | gah, can't remove leasesC - we can't run transactions against an unknown collection. | 02:50 |
| babbageclunk | wallyworld: ^ | 02:50 |
| wallyworld | which txns? | 02:50 |
| babbageclunk | the ones in MigrateLeasesToGlobalTime | 02:50 |
| wallyworld | can we chck if the collection exists before running hte upgrade step logic | 02:51 |
| wallyworld | if the collection is not there, no need to migrate anything | 02:51 |
| babbageclunk | yes, but it needs to be in allcollections, otherwise running a transaction fails | 02:52 |
| wallyworld | point me at the code that fails? | 02:52 |
| wallyworld | the collection will be there in older dbs right | 02:53 |
| wallyworld | and we run the upgrade step on that existing collection | 02:53 |
| babbageclunk | wallyworld: state/upgrades.go:1293 | 02:54 |
| wallyworld | babbageclunk: right, so that the top of the method we check if the collection exists | 02:54 |
| wallyworld | if it doesn't nothong to do | 02:55 |
| babbageclunk | it's not the existence of the collection - it's the entry in allCollections | 02:55 |
| wallyworld | HO? | 02:55 |
| babbageclunk | Otherwise the tests (and the upgrade step) fail with forbidden transaction: references unknown collection "leases" | 02:56 |
| babbageclunk | sure, in stdup | 02:56 |
| tlm | jjkkjj | 03:08 |
| tlm | wallyworld: got 5 minutes for HO ? | 03:18 |
| thumper | https://github.com/juju/juju/pull/11584 - cleanup for some future work | 03:19 |
| wallyworld | tlm: sure | 03:20 |
| hpidcock | thumper: lgtm | 03:23 |
| thumper | hpidcock: thanks | 03:46 |
| thumper | https://github.com/juju/juju/pull/11585 - for a bug that had been assigned to me for over a year | 03:46 |
| thumper | or two | 03:46 |
| hpidcock | lgtm | 03:47 |
| thumper | hpidcock: thanks again | 03:48 |
| tlm | hey wallyworld you didn't push a tag for juju/description | 03:57 |
| tlm | still have it? Not sure I have push access to that repo | 03:57 |
| wallyworld | ah, it was addd to the branch for which i created a PR | 03:57 |
| wallyworld | i'll add | 03:57 |
| tlm | ta | 03:57 |
| wallyworld | tlm: try now | 03:58 |
| tlm | works cheers wallyworld | 04:01 |
| tlm | wallyworld: I think go mod wants tags in the form of v2.0.1 hpidcock can you confirm ? | 04:03 |
| hpidcock | tlm depends on the package | 04:03 |
| wallyworld | tlm: the previous tag was 2.0.0 | 04:03 |
| tlm | hmmm go mod is not liking life | 04:06 |
| wallyworld | tlm: i added v2.0.1 | 04:07 |
| wallyworld | does that help? | 04:07 |
| tlm | changed the error but hasn't helped | 04:08 |
| tlm | will dig into it | 04:08 |
| tlm | can you remove the v tag wallyworld ? | 04:09 |
| wallyworld | ok | 04:09 |
| wallyworld | done | 04:12 |
| wallyworld | babbageclunk: you also removing store.Refresh() ? | 04:32 |
| wallyworld | tlm: looks good after the go.mod issue is sorted | 05:17 |
| tlm | wallyworld: go.mod issue should be sorted now | 05:18 |
| tlm | ? | 05:18 |
| wallyworld | tlm: ah ok, sorry i was probably looking at an outdated diff | 05:18 |
| babbageclunk | wallyworld: yeah, I will - just working out how I can make the dummy store do autoexpiry | 05:26 |
| wallyworld | ah joy | 05:27 |
| wallyworld | babbageclunk: i have the unit->dead revocation fully working :-D | 05:27 |
| babbageclunk | ooh nice | 05:27 |
| wallyworld | kelvinliu: tlm: so i think your stuff is good to go; i have release notes organised; as soon as things land wash through jenkins, i'll kick off rc2. any issues i'm missing? | 05:30 |
| kelvinliu | wallyworld: ho? | 05:31 |
| wallyworld | ok | 05:31 |
| wallyworld | hpidcock: you free to jump into standup? | 05:41 |
| hpidcock | wallyworld: sure | 05:47 |
| wallyworld | hpidcock: kelvinliu: sorry, hit button too soon | 06:07 |
| kelvinliu | nws | 06:07 |
| hpidcock | wallyworld: sure | 06:07 |
| babbageclunk | wallyworld: hey, I'm just going to hit merge on my pr so yours doesn't need to hold up - I can always put the extra store tidy-ups in after. | 06:19 |
| babbageclunk | that was poorly worded but I think you can probably see what I'm getting at | 06:19 |
| wallyworld | babbageclunk: whoohoo, tyvm | 06:20 |
| achilleasa | manadart: I have landed the []InterfaceInfo -> InterfaceInfos PR on 2.8 and dev; make sure to rebase if working with those types to avoid conflicts | 08:21 |
| manadart | achilleasa: Yep, rebased already. | 08:21 |
| achilleasa | manadart: have a question on 11579 | 08:26 |
| manadart | achilleasa: HO? | 08:27 |
| achilleasa | omw | 08:27 |
| manadart | achilleasa: Pushed change and replied to your comment. | 09:21 |
| achilleasa | looking | 09:21 |
| manadart | achilleasa: Simple one: https://github.com/juju/juju/pull/11587 | 11:00 |
| manadart | achilleasa: Thanks. Also need a tick on this backport of the earlier one: https://github.com/juju/juju/pull/11588 | 11:10 |
| achilleasa | manadart: done | 11:10 |
| manadart | achilleasa: Ta. | 11:11 |
| manadart | achilleasa: Trivial forward-merge https://github.com/juju/juju/pull/11589. | 13:29 |
| achilleasa | manadart: trade you for https://github.com/juju/juju/pull/11590 | 13:39 |
| manadart | achilleasa: Deal. | 13:40 |
| === narindergupta is now known as narinderguptamac | ||
| === grumble is now known as rawr | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!