Temporary failures during server deletion leak server resources

[mutability]

While running an uninstall workflow:

2015-02-05T19:02:51 CFY <testrhino> [rhino_db_host_61ad7] Deleting node
2015-02-05T19:02:51 CFY <testrhino> [rhino_db_host_61ad7] Deleting node
2015-02-05T19:02:52 CFY <testrhino> [rhino_db_host_61ad7.delete] Sending task 'nova_plugin.server.delete'
2015-02-05T19:02:52 CFY <testrhino> [rhino_db_host_61ad7.delete] Task started 'nova_plugin.server.delete'
2015-02-05T19:02:57 CFY <testrhino> [rhino_db_host_61ad7.delete] Task failed 'nova_plugin.server.delete' -> RecoverableError("ConnectionError: ('Connection aborted.', gaierror(-3, 'Temporary failure in name resolution'))",) [attempt 1]

The failure is ignored and not retried, and now that server has been leaked.

[mutability]

and subsequently things like subnet deletion also fail because the server is still present, so they are also leaked.

[uric]

@mutability can you please post in the users mailing list first? Also, once you've confirmed there that it's indeed a bug / issue, please post in our Jira. See more details here.

[mutability]

Sent a copy to the list. It's definitely a bug that the plugin/workflow leak cloud resources, surely?