I'm having difficulties explaining the enormous memory consumption on a PI Data Archive server.
The server is running on Windows Server 2008 R2 with PI Data Archive 2015 SP1.
12 GB memory
Collecting 50 HZ data on approx 150 tags
Only 400 days of data online (in archives)
It apears that the server manages to allocate 5-7 memory shortly after reboot as we have a PI Perfmon interface running that never picks up more than 5-7 memory available (again, the server has 12 GB of memory).
Secondly, after some time all memory gets consumed. The client has a service running that collects data trought the entire history meaning all archives is read (and partly loaded in the memory) at some point. But once this is finished it never releases the memory again which slows down the system.
Both spikes in available memory is caused by restarting PI Archive subsystem
Reboot of server (the service that collects history has runned one time at the beginning. Note that's 5 days ago)
Memory consumption last 8 days.
Memory consumption *-16d to *-8d (tags for pool and comitted wasn't created at this point)
I've tried to use several tools to find where the memory is consumed
Task mngr (ordered by peak) as you can see, no service is claiming to use this much memory:
Resource monitoring, same storry. Note that close to nothing is cached or available.
Using Sysinternals RamMap i've located that 5 GB is "Mapped file" and 5 GB is "Paged pool" but once I digg into the details again I can't really explain the usage.
Looking at the "File summary" I find that most of the archives is partly active, I have 60 archives that uses more than 50 MB. So when I take 60MB x 60 archives I have 3,6 gb memory consumed + the primary archvie 826 MB = 4,4 GB memory. After that I have a few archives between 10 and 42 MB of memory consumed, meaning I've located the "Mapped file" consumption of approx 5 GB memory. But where didd the "Paged pool" end up? and why isn't the server releasing memory over time? Any suggestions to find where my last 5,4 GB of memory is gone?