[01:54] <thumper> wallyworld: can we chat now?
[01:54] <thumper> start a bit early?
[01:54] <thumper> I have a feeling I'm going to have to run a bit early too
[03:04] <babbageclunk> Does anyone know why the pre-push script was deleted?
[04:41] <wallyworld> babbageclunk: yeah, i was wondering why my commits were going so fast
[12:02] <jam> manadart: so it is worker/machiner that transitions from Dying to Dead, which then tells worker/provisioner that it can StopInstance
[12:03] <jam> and machiner is calling EnsureDead which gives us a chance to make sure we don't go to Dead as long as we HaveVote.
[12:04] <jam> we'd still need something to remove this machine from votingmachineids and machineids.
[12:10] <manadart> jam: Could we make votingmachineids derived for a start?
[12:11] <jam> manadart: so 'votingmachineids' is supposed to be the denormalization of... WantsVote ?
[12:25] <jam> manadart: afaict, 'machineids' is denormalizing JobManageModel, and 'votingmachineids' is denormalizing !novote
[12:38] <manadart> jam: Do we gain more from the denormalisation than we hurt when needing to tear-down?
[12:39] <manadart> jam externalreality: https://github.com/juju/juju/pull/8517 accommodates feedback to-date including new error messages following our discussion.
[12:40] <jam> manadart: so I think at least one of those denormalizations is useful, as it means peergrouper can watch for changes in that set, rather than having to watch all possible machines
[12:40] <jam> but that's probably only 'machineids'
[12:41] <jam> manadart: votingmachineids doesn't seem to be valuable (to me)
[12:44] <externalreality> manadart, that error message doesn't seem to be too different from what is was before.
[12:50] <manadart> invalid config value for "juju-mgmt-space": machines with no addresses in space "mgmt-space": "0"
[12:50] <manadart> versus
[12:50] <manadart> invalid config "juju-mgmt-space"="mgmt-space": machines with no addresses in this space: 0
[12:51] <manadart> I didn't do the <count> machines have no...
[12:51] <manadart> You have the singular vs plural thing with that.
[12:53] <externalreality> manadart, "there are controller machines that are not in space <space>: 0, 1, 2"
[12:53] <externalreality> That would leave little room for any confusion or ambiguity, no?
[12:54] <externalreality> Its not a hill to die on however, LGTM
[14:42] <jam> balloons: I put together an ugly hack, but it lets me bootstrap on bionic with mongodb-server-core.
[14:42] <jam> balloons: some of it may survive, but I thought you might be interested in testing it.
[14:42] <jam> I sent you an email about it.
[14:44] <balloons> jam, awesome, thanks
[14:58] <hml> one line review anyone?  https://github.com/juju/juju/pull/8522
[14:59] <manadart> hml: Approved.
[15:01] <hml> manadart: ty!
[17:31] <cmars> hi, can i get a review of https://github.com/juju/juju/pull/8513 ?
[17:31] <cmars> balloons: ^^ ?
[17:31] <balloons> it's on the list cmars
[17:32] <cmars> balloons: awesome, thanks!
[22:39] <thumper> I'm happy now that the team calls have changed times, I can now make lunch time gym class
[22:39]  * thumper rolls
[22:57] <anastasiamac_> anyone keen to review https://github.com/juju/juju/pull/8519
[22:58] <anastasiamac_> thumper: babbageclunk: wallyworld ^^
[22:59] <babbageclunk> anastasiamac_: yup, looking
[23:00] <anastasiamac_> babbageclunk: my hero! thnx \o/
[23:30] <anastasiamac_> babbageclunk: i agree with difficulty of catching everywhere -  we just need to be cautious when reviewing changes to provider/adding new functionality
[23:30] <anastasiamac_> babbageclunk: adding the conversion on the other side does not help however...
[23:31] <anastasiamac_> babbageclunk: we wrap errors on our side and if we'd wrap "cred invalid" error, we'd lose its type anyway...
[23:31]  * anastasiamac_ commenting on PR
[23:31] <babbageclunk> anastasiamac_: I think you can't add it higher up (since you need to look for it in some places in the code), but maybe we could add something to the library - everything ends up going through the query method (I think after not very much reading).
[23:33] <anastasiamac_> babbageclunk: again, the problem is that we wrap errors in our code (after the library calls) and will lose the type.... hence, why StartInstance was not an easy change -  we sometimes expect a typed ZoneIndependent error but for crednetial we actually want it to b a different type...
[23:34] <babbageclunk> anastasiamac_: ? I'm suggesting you convert the problem ones before that though.
[23:34] <anastasiamac_> babbageclunk: can we talk in a-team standup?..
[23:34] <babbageclunk> yes!