[AMBARI-25614] DB purge logic is wrong

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Won't Fix
• Assignee: Unassigned
• Reporter: ramkrishna.s.vasudevan
• Created: 03/Feb/21 09:10
• Updated: 04/Feb/21 18:01
• Resolved: 04/Feb/21 18:01

Description

While purging the DB, we get the

      // find task ids using request stage ids
      Set<Long> taskIds = hostRoleCommandDAO.findTaskIdsByRequestStageIds(requestStageIds);
      // find host task ids, to find related host requests and also to remove needed host tasks
      final Set<Long> hostTaskIds = findHostTaskIds(taskIds);
      // find host request ids by host task ids to remove later needed host requests
      final Set<Long> hostRequestIds = findHostRequestIds(hostTaskIds);
      final Set<Long> topologyRequestIds = findTopologyRequestIds(hostRequestIds);

The hostRequestIds are already got from the topology_host_task table. Those hostRequestIds are now used to find the topologyrequestIds. Internally those should be got from the topology_logical_request table. But we instead again go to the topology_host_task table.

Here what evern toplogyRequestIDs we get do not match with the actual Topology_requests. Hence the delete fails saying no such requestID found.