Skip to Content
Jun 01, 2009 at 08:08 PM

Memory Allocation Observation



Recently experienced the situation where one of our AIX application servers went into memory distress. The server has been in production for a significant amount of time and there are 4 other servers that share the production load and none have ever experienced this and it is a first for this particular server as well.

Via SM04 observed that a user using only a single session was using 32538 Megabytes of memory. Without looking any further I went to SM50 and cacelled without core the workprocess that was executing this thread.The memory wasnearly instantly cleared up and the server returned to normal operation

My question is with the memory parameter as defined below how could this have happened? I would have thought that long before it got to this point it would have shortdumped with TSV_TNEW_PAGE_ALLOC_FAILED or something similar.

Thanks for any insight,


Other Information:


Physical Memory = 12Gb

Paging Space 20Gb

Memory Parameters:

ztta/roll_area 6500000

ztta/roll_first 1

ztta/short_area 2000000

rdisp/ROLL_SHM 16384

rdisp/PG_SHM 16384

rdisp/PG_LOCAL 150

em/initial_size_MB 16383

em/blocksize_KB 4096

em/address_space_MB 4092

ztta/roll_extension 2000000000

abap/heap_area_dia 10000000

abap/heap_area_nondia 10000000

abap/heap_area_total 2000000000

abap/heaplimit 20000000