Hi,
may someone has a similar Problem or can give me a helping Hand.
I'm having a 9Node Cluster ( Windows 2012 R2 , fully Patched from RTM Version on )
The Cluster is connected to a SAN which is an Equallogic ( Firmware 6.11). HitKit Driver 4.7.1
The System running clean and without any Event Logs until the Page Pool Usage of the Nodes turn over 15 GB of Paged pool Memory.
I monitore that via a Performance Counter ( \Memory\Pool Paged Bytes)otherwise I have no indication by for example the Process View, the Calculated Sum on the Overview ( TaskManager / Memory ) but shows on the contraditory the 15GB an Memory Usage.
I downloaded the RAMMap Tool from Sysinternals but that doesnt show me any possible Information for the information , acutally the Information differs from RAMMap to the Performance Counter ( RAMMap = 251MB, Performance Counter = 15GB )
Just to make that point clear , there is No Information beside the Sum of Page File on the System Page and the PerfCounter that so much Page Pool Memory is used by any Process . Therefor it is hidden which process "needs" it.
I tried many thinks , like search for a Memory Hole in the Drivers , Hyper-V Stack , Equallogic and hoped for resolution with any new Patch which had a "Memory Leak" Information to them , but so far none has made a difference , maybe someone has the same Problem.
My Cluster has routher large VMs on it compared to the Numa Seize ( VM 4-128GB, on DELL R720 ( 2x CPU (8C) 384GB RAM )
( The VMs have also a High IO Usage , as there RAM is used as a RAM Database ( zb. Mongo DB ) )
Regards
M