Because the server acts snappy when MEM free runs low... Also the server is quite old and running on HDDs, so swapping is really slow...

Gabriel Gheorghiu said:

snappy

Normally snappy implies fast and quickly responding, perhaps too fast.  I suspect that wasn't what you were trying to say from the tone of the rest of what you wrote.

For an iPrint only server, the question is what humanly perceivable performance issues are there that can be directly attributed to the server?  Also would be worth checking the drive system to make sure it doesn't have any issues (such as a drive down) as all print jobs do get written to disk as part of printing, so any slow-downs there would be a notable impact way beyond anything those memory numbers you shared would indicate.

Have you used ganglia to see the memory trends?  http://serverIP/gweb   

I have a client of about the same size, with their iPrint vm working just fine with 4GB of ram, even with the charge-back mechanisms in place.

Hey Andy,

lol , no that wasnt what i ment and thanks for clearing that up. Now it makes totally sense that other people reacts weird when i used the word like i did...

Its not an iPrint only server... Its the main server hosting NDS, NSS, DHCP, DNS, LDAP and iPrint Advanced...

I didnt check the ganglia service but i will...

I think i will move the iPrint Service to its own appliance

Keeping header on top, with :

ps -eo pmem,pcpu,pid,user,args --sort -pmem|head -15

Ahh,  so much cleaner on those systems with pids that use more than 9.9% MEM, 
no difference otherwise.  So worth keeping this in my kit.  Thank You

A note for others.  the data is like the body of
   top {enter} c M
but cleaner, scriptable, and doesn't shift as it updates.