> I wouldn't worry too much about figuring out all the parameters to > StartInstance *in this branch*. Better to get something landed that covers > the very basic point (find a node and start it) and then look at ticking all > the other boxes. It will make review easier and it'll produce more focused > testing, too. > > On another note, I wouldn't just abort StopInstances() on failure. I think > the equivalent of what the ec2 provider does is: stop each of the nodes, > ignoring errors. Then loop over them again to stop them once more, but this > time check for errors. That ensures that even if something fails halfway > through, you will make an attempt to tell each of the nodes that it should > shut down. > > That last iteration can ignore "node not found" errors. Not as important as > with ec2 though, because MAAS nodes don't just disappear entirely merely > because you shut them down.
Good point, I've used a different strategy: iterate over all the instances and issue the stop signal, collecting all the error, then return the first non-nil error (same as what the ec2 provider does). -- https://code.launchpad.net/~rvb/juju-core/mpv-start-stop/+merge/147167 Your team MAAS Maintainers is requested to review the proposed merge of lp:~rvb/juju-core/mpv-start-stop into lp:~maas-maintainers/juju-core/maas-provider-skeleton. _______________________________________________ Mailing list: https://launchpad.net/~launchpad-reviewers Post to : [email protected] Unsubscribe : https://launchpad.net/~launchpad-reviewers More help : https://help.launchpad.net/ListHelp
