[17:00] <hbogert> I'm probably misunderstanding `juju migrate` because I don't understand how this would work for storage that the model depends on. Is model migration only for stateless applications?
[17:20] <rick_h> hbogert: so the running workloads don't migrate
[17:21] <rick_h> hbogert: it's just the controller server moving control/tracking/communication to a new controller server
[17:21] <rick_h> hbogert: the running applications don't change at all. The agents running on those machines are updated to start talking to the new controller as its master and commander of all things
[19:24] <EdS> hey hey Juju folks :)
[19:24] <EdS> I have a failed machine - what's the best way to replace it now I've fixed the problem? The hdd has been replaced but all other hardware was fine. This is in MAAS...
[19:37] <EdS> do I need to juju remove-machine then add-machine?
[20:26] <hbogert> EdS: just learning myself, but unless removing the applications or their units first is not a option in your case, then yes,  remove-machine should work for you
[20:27] <EdS> hbogert: I have just managed this. The ironic thing is the wiki containing my notes was inaccessible because of the problem! *headdesk*
[20:28] <EdS> I had to "juju remove-unit <affected unit>" then "juju add machine" to get MAAS to deploy a fresh machine to replace it, then "juju add-unit <unit>" to get back the missing services.
[20:29] <EdS> not too painful, but painful when one's own docs are torn from the hand ;)
[20:35] <hbogert> ahh yes, well I have similar sins in my setup 8)
[20:36] <EdS> yeah - I think that sort of thing is common. Good night all - may your disks spin smoothly ;)
[20:53] <hbogert> I wish to migrate a single juju charm application, i.e., easyrsa, to another node. It's currently deployed on through MAAS, how would I approach this?