Assuming this is related to your memory leak post from a few days ago, the first thing to look at is that if you have an ever-widening delta between ‘.net/bytes in all heaps’ and ‘process/private bytes’ then I think an unmanaged memory leak is pretty much confirmed.
From: Greg Keogh Sent: Friday, January 30, 2015 8:37 AM To: ozDotNet As usual, I asked my cat, and he said to just click every perfmon node and button to see if you find anything useful. It turns out the "trick" is to click the Performance Monitor node (which shows current activity), then click an indistinct small blue button with tip "View Log Data (Ctrl+L)" from where you can add log (*.blg) files, which is what I collected overnight. Then you can pick the counters to chart and see history. Now I just have to figure out how to interpret things like Pool (Non)Paged Bytes and Free PTEs to determine if something is suspicious or not. Greg K On 30 January 2015 at 11:26, Greg Keogh <[email protected]> wrote: Folks, I generated a perfmon Data Collector Set of several important counters in 15 second intervals for 24 hours and I have the generated files and can see a "report" in the perfmon UI. The report simply shows the Mean, Minimum and Maximum of the selected counters. Perhaps I misunderstand what perfmon is supposed to do for me, but I expected to be able to generate charts from the 2.6MB of collected samples, but web searches and help doesn't provide any useful hints on how to do this (or if it's possible). I expected to look at history charts and identify peaks of memory and CPU usage so I could attempt to match this to our application logs. I must be missing some vital clue about what I can do with my Collector Set. Any hints please? Greg K
