Subject FB Memory Use
Author Alan McDonald
Following on from a previous thread on memory use:
Can someone explain to me how these numbers relate, which numbers are
important / not important to me and why perhaps they vary so much.

Server example as a snapshot today with 12 simultaneous connections:
Task Manager:
Mem Usage: 114,528K
Peak Mem Usage: 185,344K
VM Size: 159,816K

isc_database_info() returns
CurrentMemory: 68,881K
MaxMemory: 116,125K

New connections add another approx 2Mb of RAM usage and when they disconnect
the ram drops back - I can reliably observe the recovery in ram usage.

I am trying to find why over the last couple of days I have experienced
(twice) a case of an exception (cannot allocate memory from OS) where the
server MemUsage in Task manager was reporting over 1Gb of ram and
connections were being refused.

I was monitoring the MON tables continusously on both occasions over a
period of several hours.
I am trying to discover if:
1. it was caused by continuous monitoring of the MON tables
2. if it is something in the monitoring software causing this buildup, or
3. something else like an artifact of the .NET driver used in the monitoring
Since I ceased monitoring the servers have been behaving quite normally
again (in terms of memory use).
But I would like to start by more fully understanding how to interpret these
ram use numbers.

Does a buildup of transactions requiring barbage collection add to ram use
Does the action of a sweep add considerably to ram use?

 Alan McDonald