Glance/Memory

[Junior C.]

Glance Mem Util is at 96%
How can I tell what is causing memory to hit 96 - 98%

Thanks

Junior

[Brian M. Fisher]

This is probably due to your buffer cache settings. Within Glance, check Reports / Memory Info / Memory Report. This will show you how much memory is being used for System vs. Buffer Cache vs, User.

Brian
<*(((>< er

[Junior C.]

Phys mem: 128mb
User mem: 89.2mb
sys mem: 16.3mb
buf mem: 16.3mb
total VM: 120.6mb
Active VM: 63mb
free mem: 5.1mb

[Brian M. Fisher]

To further find out how much memory each process is taking, double click on each process to find RSS (resident memory) and VSS (virtual memory)

Brian
<*(((>< er

[Jim Welch]

You can also run gpm and go to the process list and sort processes by the rss value to see which ones are using the most ram. (make sure your filters and sort fields are set (Configure|Filters) (Configure|Sort Fields)

[Bill Hassell]

It's important to understand that 100% memory usage in Unix is not a problem but quite desirable. As a virtual memory operating system, Unix transparently runs any amount of programs with limited RAM. All you need is enough swap space. You can easily run 2,000 megs of programs in 128 megs of RAM...perhaps slowly as the processes swap, but nevertheless, the system will run.

If swapping becomes excessive, things will run slowly...you can see this with Glance (or vmstat) looking at page outs (not page in which includes all program starts).

As mentioned before, the buffer cache may be set to a poor value. By default, it ranges from 5-50% of RAM, a rather siginifcant impact to RAM usage until processes push the limit down towards 5%. To avoid this remapping, you can set the kernel param bufpages to about 200-500 megs (the parameter is in pages of 4kb) for average systems, much higher for large RAM (multi-GB) where reads outnumber writes significantly.

To see the programs which occupy RAM sorted by size:

UNIX95= ps -e -o ruser,vsz,pid,args | sort -rnk2 | more

The UNIX95 setting turns on new options in the ps command.

[CHRIS_ANORUO]

To help with this mem. high %tage. Pls set your kernel parameters as follows:
swapmem_on=1
allocate_fs_swapmap=0
nbuf=0
bufpages=0
If nbuf and bufpages are set to zero, you can reduce dbc_max_pct to 20 to reduce the range it has to go when remapping. From your mesg. your average buf mem is 16.3mb which is 12.7% of you Phys mem.
The above kernel settings will help in the system memory allocation and usage.