[00:14] hml: did you have time for a quick hand over for next week? in standup ho? [00:14] wallyworld: sure [01:58] * menn0 is pretty sure he just fixed the mgo/txn issue which leaves txn-queue entries behind for assert-only txn.Ops [01:58] \o/ niice [01:59] before the model docs would have tons of txn-queue entries, more after each deploy [01:59] now they just have one [01:59] it was only a few new lines in mgo/txn/flusher.go [01:59] needs more testing but it's promising [02:01] menn0: since each deploy used to increase entries, we could probably have a CI test around it to ensure no future regressions?.. [02:01] anastasiamac: that's very doabel [02:01] doable [02:01] menn0: m thinking of how to test it without having long-running test [02:01] :D [02:02] anastasiamac: just deploy a few things and check the length of the txn-queue fields on some select docs [02:03] veebers: do u know of CI tests that examine db state after/as part of the run? ^^ is it an easy test to add? [02:04] * anastasiamac thinks veebers is probably recovering from anastasiamac's humor [02:06] anastasiamac: heh :-) There are no tests that I know of that introspect the db as part of the assertions [02:07] veebers: so not a simple test to add to CI?.. [02:07] anastasiamac: I'm unsure how much work would be involved. If it's possible to have a command we can run with ssh that's really handy and shouldn't be too hard [02:07] anastasiamac: it depends on what's involved with checking he db state [02:56] axw: review please: https://github.com/juju/juju/pull/7183 [02:57] looking [02:58] menn0: LGTM, ta [02:58] axw: cheers [03:41] wallyworld: if you've got nothing better to do, here's the bulk of the persistent storage changes that I've been working on: https://github.com/juju/juju/pull/7186 [03:41] and thanks for the other review [03:42] axw: ok. i didn't realise xtian had proposed his public address PR. it will conflict a bit with mine. best to land his first once approved and then i'll need to adjust [03:42] wallyworld: okey dokey [03:42] he deletes subnets stuff which i'll have to add back :-) [03:42] heh [03:42] as we will need it at some point [03:49] axw: PR 7177 looks ok at first look - there's a todo or 2 i think needs to be there , but if it lands i can rebase my PR i proposed just before and address any issues. can you take a look and i'll land if you're happy? [03:53] wallyworld: will do [03:53] ty [03:57] Bug #1662857 opened: cannot go get the source code [04:07] wallyworld: LGTM [04:07] ty [04:09] axw: i just got to pruneExistingMachineStorageAttachments and am unclear why that has been added. perhaps a comment to clarify? [04:09] wallyworld: sure, will do [05:13] nooo, using 1.8 code in my ec2 PR. maybe I can just wait for the builders to be updated... [05:36] any chance for a review on https://github.com/juju/juju/pull/7180 over the weekend? [05:43] jam: will do [07:25] wpk: hey, I believe jam and sinzui are both out. if you are blocked or need anything, I'm around for a couple of hours still [07:31] wallyworld: that pruneExistingMachineStorageAttachments call was from a previous iteration of the code, it's not necessary anymore. if the volume attachment exists when we go to attach storage, we'll error out [07:31] (so I'm removing it) [07:31] ok, i couldn't figure out what is was for [07:32] wallyworld: yep, thanks for calling that out [07:32] np [07:33] axw: xtian's branch has landed but it has borked mine and i'm working to get stuff going and hand tested on aws etc. hopefully i'll be done soon, but you never know [07:33] wallyworld: ok. well, I'm still around for a while [07:34] yeah, see how i go [07:58] axw: i'm live testing atm, but here's the PR now rebased on xtian's work (ie develop tip) https://github.com/juju/juju/pull/7184 [08:02] wallyworld: ok looking [08:08] axw: testing on aws shows the sec group ingress rule is correctly set to /32 and it works as expected [08:09] wallyworld: cool bananas. still reviewing, there's a fair bit to get through [08:09] axw: thanks [08:09] axw: yeah, sorrry. i lot is deleting. the main addition is the new watdher [08:11] axw: my idea is to get rid of bridgePrefix (from everywhere) and bridge name generation in random places and force passing of the full bridge name down from FindMissingBridges [08:14] wpk: heh, sorry that is beyond my knowledge of the juju networking code. I can help with more general thing ;) [08:15] wpk: happy to talk that through though if you want to explain - not sure I'll be helpful tho [08:24] axw: Well, currently I'm like Mr. Bean in the movie - I sit in the corner... and look at the code. [08:26] wpk: hehe. I'll finish up what I'm doing soon and see if I can understand the current bridge prefixing code, then might be able to provide useful input [08:27] axw: I'm currently working on cleaning up https://github.com/juju/juju/pull/6963 this to hit two birds with one stone and also add support for https://bugs.launchpad.net/juju/+bug/1672327 this [08:27] Bug #1672327: Too long names for bridges

[08:28] wpk: ok, cool. are you aware that I made some changes to the python version to handle multiple stanzas for the same interface? that was after the Go version was written [08:29] axw: I was not, thanks [08:29] wpk: some history: https://bugs.launchpad.net/juju/+bug/1650304 [08:29] Bug #1650304: Juju2: 'Creating container: failed to ensure LXD image: image not imported!'

[08:34] axw: kk [08:40] axw: i found 2 issues which i adds cards for. removing a relation doesn't close the ingress rule. adding the relation again a second time doesn't open the port, because it seems the watcher needs to do a better job of oading initial data [08:41] the removal issue may be pre-existing [08:41] wallyworld: mk [08:42] should be easy to fix, i'll let xtian know [08:58] wallyworld: left a bunch of comments [09:05] ty [09:30] axw: i fixed the issue with address counts and added a test [10:32] wallyworld: sorry had to go make dinner, looking [10:32] no worries at all [10:33] gotta eat === akhavr1 is now known as akhavr [10:37] wallyworld: LGTM, thanks [10:44] axw: tyvm for review [11:55] Survey: what do you think is the most unlikely character to happen in an interface name? [12:06] wpk: Trying to come up with a unique prefix? [12:30] zeestrat: no, with a separator for command line [12:30] ~ [12:31] I came up with ~ [12:35] wpk: according to https://www.wired.com/2013/08/the-rarity-of-the-ampersand/ ~ is the least used char :) === akhavr1 is now known as akhavr === admcleod is now known as not_adam === not_adam is now known as admcleod [16:19] https://www.meetup.com/SF-CockroachDB-User-Group/events/238072086/ === frankban is now known as frankban|afk