Jonas Smedegaard skrev:
| | I guess this is true, lots of different clients all behaving the same. | There should not be a need to dig too deep, it's a simple problem to | both diagnose and solve. If you have thin clients with RAM >= 32 MB RAM, | use NFS swap :-) Ahem, I think you meant the opposite there: if RAM <= 32 MB then use NFS swap :-)
I stand corrected. My apologies.
Just a thought: Maybe it would be possible to isolate the cause of the problem by somehow recognize when the NFS swap is actually used and take a snapshot of the processes running on the client, together with their footprint. This could help reveal any memory leaks or otherwise bloated processes.
I'm sure this could be done. In LTSP 4 you can read /proc so you could get the memory statistics and other pertinent easily. I guess the software used to enable this could be ported to LTSP 3, but who has access to a Red Hat Linux 7.0 machine these days ;-)
(It's a little known fact that every single thin client using Skolelinux as a server runs a stripped down version of Red Hat Linux 7.0 aka LTSP 3 ;-))
-- Med vennlig hilsen Ragnar Wisløff ------------- Life is a reach. Then you gybe.