Description
On a large, busy cluster the monitor page can be sluggish to load. This has been potentially traced back to the amount of data that needs to be gathered and to the generation of the charts.
It may be possible to give the illusion of more performance (or at least a shorter blocking wait) through some optimization techniques. Specifically, I believe we can benefit from http://developer.yahoo.com/performance/rules.html#flush although I do not know how applicable this is to Java.
Attachments
Issue Links
- is related to
-
ACCUMULO-1013 Integrate with a scalable monitoring tool
- Resolved
- relates to
-
ACCUMULO-302 long lists in the monitor pages could be improved with summaries
- Resolved