Do not fail if querying for a server throws an error in the API

Unfortunately, some serves stay in the DB with wrong metadata (probably
old servers, with some information that was not correctly migrated, and
they throw an error when requesting their information. If this happens,
there are no records being generated for the rest of the period, since
we did not catch the exception. This change fixes this.

Author: alvarolopez

caso/extract/nova.py

@@ -243,7 +243,17 @@ def extract_for_project(self, project, extract_from, extract_to):
         for usage in usages:
             # 4.1 and 4.2 Get the server if it is not yet there
             if usage["instance_id"] not in records:
-                server = nova.servers.get(usage["instance_id"])
+                try:
+                    server = nova.servers.get(usage["instance_id"])
+                except novaclient.exceptions.ClientException as e:
+                    LOG.error("Cannot get server '%s' from the Nova API, "
+                              "probably because it is an old VM that whose "
+                              "metadata is wrong in the DB. There will be no "
+                              "record generated for this VM. " %
+                              usage["instance_id"])
+                    LOG.exception(e)
+
+                    continue
 
                 server_start = self._get_server_start(server)
                 if server_start > extract_to: