ad_response probe slowing down over time

Discussion created by christopher.harris on Nov 3, 2009
Latest reply on Nov 4, 2009 by keith_k
Has anyone noticed a problem with the profiles in the AD response probe showing longer latencies over time?

After decreasing the sampling rate from 10 minutes to 1 minute, the response time gradually increased from an average of 100ms to 300ms over about 30 days.

No applications were complaining and the DCs aren't under any increased load according to CPU, Disk, Mem.

I made no changes to the DCs, but rebooted the Nimbus Hub server and the average response times dropped back to about 100ms.

The one anomaly I noticed prior to rebooting the Nimbus server was that the LSASS.exe process was consuming a whopping 9GB of memory (Private Working Set). Unfortunately, I wasn't monitoring memory consumption by lsass previously, so I can't tell if the increased latency correlated with an increase in memory consumption by lsass, but it seems likely.

If anyone has seen this and figured out a way to prevent the problem short of rebooting the Hub server I would appreciate any advice you have.