Hive Server interruptions should have consistent error message and backend should return HTTP code 503

Description

Currently, Catalog page shows error "Unable to list tables for catalog 'null' and schema 'null'" when hive server is down. This is misleading and doesn't help the user understand the error from the back end is that the hive server may be down and that they need to enlist the help of an admin.

To repro:
1. In a normally running Kylo go to Catalog page and choose Hive
2. You will get a list of schemas
3. go to command line and stop hive server `service hive-server2 stop` in CDH
4. Empty cache and hard reload UI so the cache of schemas is cleared
5. Notice the following error:

Note: this should be handled in much the same way that Kylo handles the situation where NiFi is down (see ticket: ). If a downstream service is unavailable a 503 error should be thrown. Any attempt to contact thrift server that fails should throw a common exception e.g. HiveServerConnectionException and a ConnectionExceptionMapper (example NifiConnectionExceptionMapper) should interpret the error into a user friendly message.

Environment

None
Done

Assignee

Tim Harsch

Reporter

Tim Harsch

Labels

None

Reviewer

None

Story point estimate

3

Story Points

3

Components

Sprint

None

Fix versions

Affects versions

Priority

Medium