[OOZIE-1716] Add a way to stream the logs for only the last x hours

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: trunk
• Assignee: Purshotam Shah
• Reporter: Robert Kanter
• Created: 27/Feb/14 21:04
• Updated: 17/Jul/14 23:42
• Resolved: 17/Jul/14 23:07

Description

When a user has a busy cluster and long running coordinator jobs, streaming the logs can be really slow. In this situation, the user is typically only interested in the latest logs and doesn't care about logs from a while ago. So, we can speed this up by adding a parameter or something to the log streaming API to only fetch the logs from the last x hours. The log files roll over every hour, so this should be pretty straightforward. We can also prepend a message to the logs streamed back mentioning that only the last x hours of logs are included. The existing way of getting all of the logs should remain unaffected and still work.

Issue Links

• duplicates: OOZIE-1737 Oozie log streaming is slow