[UIMA-5561] DUCC Web Server (WS) queries DB too often

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.2.2-Ducc
• Component/s: DUCC
• Assignee: Lou DeGenaro
• Reporter: Lou DeGenaro
• Created: 12/Sep/17 12:31
• Updated: 12/Sep/17 13:26
• Resolved: 12/Sep/17 13:26

Description

In order for the WS to determine if the DB is up, it reads the checkpoint dataset. However, this operation is performed for each individual user request for the daemons page. If say Zabbix and others are making such requests this could amount to a plethora of reads of the checkpoint dataset.

Evidence is a slew of entries like so:

11 Sep 2017 13:37:25,791 INFO DB.HistoryManagerDb - 365696 restore Restored class org.apache.uima.ducc.transport.event.common.DuccWorkJob
11 Sep 2017 13:37:25,791 INFO DB.HistoryManagerDb - 365692 restore Restored class org.apache.uima.ducc.transport.event.common.DuccWorkJob
11 Sep 2017 13:37:25,791 INFO DB.HistoryManagerDb - 366293 restore Restored class org.apache.uima.ducc.transport.event.common.DuccWorkJob
11 Sep 2017 13:37:25,791 INFO DB.HistoryManagerDb - 365794 restore Restored class org.apache.uima.ducc.transport.event.common.DuccWorkJob

A cache is introduced. If DB checkpoint dataset access has been attempted within the last minute, we return the last status of: Up or Down. Otherwise WS attempts to read the checkpoint dataset and resets the status sets the time of last access attempt to the current time.