[00:39] <blr> cjwatson: thread was reasonably constructive actually
[00:40] <blr> I am surprised.
[00:40] <cjwatson> oh, the reddit one?  yeah
[00:40]  * blr heads off for lunch
[02:13] <blr> wgrant: so I destroyed the juju environment and re-ran the spec and turnip is successfully deployed because fuck computers.
[02:13] <wgrant> blr: Heh, helpful.
[02:14] <blr> no floating ip yet however
[02:15] <blr> do we want to attend to that later?
[02:15] <wgrant> I think it's important to get it sensibly deployable first, before investigating floating IPs.
[02:17] <wgrant> So ensure it's working reliably, and not just accidentally working once.
[02:17] <wgrant> Get basenode, LS, ksplice, etc. known to be working.
[02:17] <blr> yep
[02:46] <blr> wgrant: sudo uptrack-uname -r reporting, assuming ksplice is working.
[02:48] <blr> wgrant: a few nrpe failures I need to fix
[02:48] <blr> but everything appears to be deploying nicely.
[02:49] <blr> only issue remaining is the ubuntu bootstrap node - suspect it will work on wendigo, but I would feel more comfortable if fetching that charm didn't fail against the local provider.
[02:52] <wgrant> blr: What's the local error?
[02:53] <blr> wgrant: branch permissions/NotABranch
[02:53] <blr> which seems very weird
[02:54] <blr> if you have a moment to test it, add this following to collect and run the spec against the local provider
[02:55] <blr> wgrant: see ops
[06:25] <wgrant> blr: I'm trying to speed up juju deployment, and turnip seems to run apt-get update three times for a deploy without any relations.
[06:25] <wgrant> Do you have any ideas on cutting that down a bit?
[06:26] <blr> wgrant: set the following in ~/.juju/environments:
[06:26] <blr> enable-os-upgrade: false
[06:27] <wgrant> Oh no, this is run by the charm's hooks, AFAICT.
[06:27] <wgrant> Like, each time the hook fires it tries to reinstall python-virtualenv etc.
[06:27] <blr> hmm
[06:30] <blr> I'm not certain, perhaps I can have a look at that tomorrow.
[06:34] <blr> a 30s wait between adding services and relations unfortunately didn't work.
[06:37] <wgrant> Hmm.
[06:38] <wgrant> I need to look at how block-storage-broker and storage work.
[06:38] <wgrant> blr: Can I switch juju into some debug mode so it will tell me when it fires hooks, when hooks complete, etc.?
[06:38] <wgrant> Like, I can see hook output, so I can try to guess.
[06:39] <wgrant> But I can't see what it thinks it's trying to do.
[06:39] <blr> I might revert that last commit - should probably be making these changed directly on wendigo and testing, just ackward working there
[06:39] <blr> yep there's hook debugging via tmux
[06:39] <blr> https://jujucharms.com/docs/authors-hook-debug
[06:40] <blr> hmm at this point I may be doing more damage than good... will have a fresh look in the morning
[06:40] <wgrant> Yeah, I've used that, but I just really want it to log when it invokes a hook, which arguments it used, and when the hook finished.
[06:40] <wgrant> Not trap them, exactly.
[06:40] <wgrant> Heh, it is late.
[06:40] <wgrant> Night!
[06:40] <wgrant> We are close.
[06:41] <blr> yes! shame about the false positive with the inverted relationships hah
[06:41] <blr> will get b-s-b sorted tomorrow
[06:41] <wgrant> Well, that's one of the problems, at least!
[06:41] <blr> night