> Memory-mapped files will account for both virtual and, to the extent > that they are resident in memory, to the resident size of the process. > However, your graph:
Correcting myself in the interest of providing correct information, this doesn't seem to be true - at least not always. I don't know whether it is a matter of different kernel versions or something else. I have definitely seen Python processes accessing CDB:s having a high resident amount and seen it creep down again as it is replaced by other data being paged in, but some Cassandras now running on 2.6.32 don't exhibit this effect. -- / Peter Schuller
