AnsweredAssumed Answered

Collector(s) High Harvest Duration

Question asked by JKW on Mar 18, 2013
Latest reply on Mar 29, 2013 by JKW
I am running version 9.1.4. One MoM, 4 Collectors. I'm seeing high harvest duration times, up to 15 seconds. They occur every hour. During the rest of the hour Duration times are less than a second.

So every hour on each collector, I get this for a couple minutes:

JVM is not pegged. However, GC increases at the time (Not overloaded)
CPU reaches 96% for about a minute during the harvest.
Havest capacity spikes to 86 percent
Number of Metric Data Queries per interval spikes (No one's using it)
Under "Query" I'm getting a spike in Data Points Returned Per Interval
Also, "Metrics Read from Disk Per Interval" Spikes
Total Index Inseration Duration Per Interval(ms) Spikes
Harvest Duration reaches 15 seconds.

It looks like the EM is doing a query of the SmartStor Database each hour. SmartStor Duration looks fine; however, Harvest Duration goes to 15 seconds.
I'm curious about the Total Index Inseration Duration. Is Introscope querying the Smartstor database and rebuidling an index on it each hour?

Anyone else seeing something similiar?

JW

Outcomes