[06:40] <jam> morning all
[12:17] <achilleasa> manadart_: can you point me to a test where we set up a real space topology?
[12:19] <achilleasa> I have a *state.Machine and want to add some subnets and add addresses in each one to the machine
[12:22] <manadart_> achilleasa: As in "integration" test with a real Mongo? Not sure we have such an example.
[12:22] <achilleasa> yeah...
[12:22] <achilleasa> :-(
[12:23] <achilleasa> I should just mock the call and move on (it's in the api/client bits)
[12:23] <achilleasa> otherwise it's add subnets, add LLDs, add addresses...
[12:23] <manadart_> achilleasa: See the tests for `GetNetworkInfoForSpaces` that's pretty close.
[12:24] <manadart_> achilleasa: `BridgePolicy` tests should have mock examples.
[12:24] <achilleasa> given that the test is essentially testing the unserialization logic (I have separate tests for the facade side) I will just inject a payload in the response... pretty sure I 've seen a few tests like that
[13:57] <achilleasa> manadart_: can you also look into this one as well? https://github.com/juju/juju/pull/11882 (it looks large because I had to add mock bits in various places for the tests)
[13:59] <manadart_> achilleasa: Yep. Gimme a few.
[14:18] <hml> achilleasa:  can you please review: https://github.com/juju/juju/pull/11883
[14:20] <achilleasa> hml: looking
[14:26] <achilleasa> hml: btw, any idea why is RefreshClientSuite.TestLiveRefreshManyRequest trying to connect to api.snapcraft.io? Shouldn't we be mocking that in the test? Got a test failure due to a DNS timeout
[14:33] <hml> achilleasa:  i’m not sure what is going on with that test.  thumper pointed it out too last night.
[14:34] <hml> achilleasa:  stickup kid wrote the test to be run both with mocks and live.  perhaps were running the live version sometimes?
[14:38] <hml> it’s interesting tha TestLiveInstallRequest is passing
[14:38] <hml> i meant TestLiveRefreshRequest
[14:38] <achilleasa> we might be using build tags for that
[14:38] <achilleasa> (for the live ones I mean)
[14:39] <hml> achilleasa:  the other thing is that the api being pinged is under development so something might have changed.  dunno yet.
[14:40] <achilleasa> wouldn't a live test be part of the bash suite?
[15:03] <manadart_> achilleasa: Left a few suggestions on your patch. I have to head home now, but I will follow up in the morning.
[15:03] <achilleasa> manadart_: thanks!
[15:17] <achilleasa> hml: the QA steps don't seem to completely match the expected output: https://pastebin.canonical.com/p/WBCWQtDCVy/
[15:18] <achilleasa> did the migration happen too quick?
[15:22] <hml> achilleasa:  looking
[15:23] <hml> achilleasa:  what command did you run for the output at line 1?
[15:25] <achilleasa> it was the 'juju migrate seven migrate-from ; juju deploy mysql'
[15:25] <hml> achilleasa:  did you run the juju switch migrate-to:admin/seven?
[15:28] <hml> for line #9, you need to run quickly after the migrate, perhaps that’s why you get a slightly different message there
[15:28] <hml> i may have missed a juju switch in there at the end
[15:29] <hml> looks like before the upgrade-controller, need a juju switch
[15:32] <hml> did a few tweeks to the qa instructions
[15:51] <achilleasa> hml: trying again. brb
[16:41] <achilleasa> hml: we really need to do something about those bogus error messages :D
[16:51] <achilleasa> hml: hmmm shouldn't it stop after complaining that the precheck failed?
[16:52] <achilleasa> it actually went ahead with the migration even though it complained... then it just erros that the model has already been migrated