scheduler driver live_migration shared storage issues

[Chris Behrens]

_live_migration_common_check tries to catch exception.NotFileFound when trying to do rpc.call()s... but something feels wrong here.

I'm refactoring scheduler tests and found this, but I don't know the intended behavior. The code around mounted_on_same_shared_storage() is confusing and doesn't seem quite right. I don't see why we raise NotFileFound only to catch it in the caller... instead of just checking the result of the calling function. Also good to note: If an exception occurs in compute manager's 'create_shared_storage_test_file', the rpc.call to it will not re-raise the original exception. rpc.common.RemoteError will end up being raised on the calling side. I could see 'create_shared_storage_test_file' failing with OSError if the dir doesn't exist, etc...

I'm noting this bug ID in the code w/ my refactor for someone else to take a look at.

[Chris Behrens (cbehrens)]

Also: in _live_migration_common_check(), when FileNotFound is caught, it is re-raised. I don't think that's an appropriate exception to raise for a scheduler call. InvalidSharedStorage() might be more appropriate.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/4856