JIRA startup causes very inefficient rebuilds of its statistics data structures

When JIRA starts up it appears to query all the issues. Unfortunately this involves n+1 queries. This is problematic when we have many thousands of issues in the db. A restart takes over 15 minutes. Instead an optimized select should be used to query the entire rowset and parse each row. This would reduce much of the wasted overhead caused by the useless transaction demarcation that wraps each select.

ReIndexing would also benefit from this optimization. I can manually perform all the select statements that retrieves all the data from the db in under 1 minute yet a reindex takes half an hour because of its inefficiencies.