Virtual Memory exhaustion on samba server [probably OT]

[Dragos]

look what I've got on the PDC earlier on today:
---
Oct  3 15:23:56 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c012b7a5
Oct  3 15:23:56 roky2 kernel: VM: killing process smbd
Oct  3 15:24:26 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c0123f51
Oct  3 15:24:26 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c012b7a5
Oct  3 15:24:26 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c0123f51
Oct  3 15:24:27 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c0123f51
Oct  3 15:24:27 roky2 kernel: VM: killing process nmbd
Oct  3 15:24:27 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c012b7a5
Oct  3 15:24:27 roky2 kernel: VM: killing process smbd
Oct  3 15:24:27 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c012b7a5
Oct  3 15:24:27 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0xf0/0) from c0133f8c
Oct  3 15:24:27 roky2 kernel: __alloc_pages: 0-order allocation failed 
(gfp=0x1d2/0) from c012b7a5
---
this is redhat 7.1 with updates, samba 2.2.1a, 256MB RAM, 280MB swap;
earlier this week I upgraded the kernel to 2.4.10 (linus tree); I red about 
some changes in the VM in the last kernel, but I didn't expected to hit me 
that hard...the last kernel I had, 2.4.7, worked fine for about 60 days with 
the same samba and no restart;

could that be a running away smbd ? when I arrived at the console it already 
killed some processes, and  swap usage was about 6 megs, nmbd wasn't running, 
so I restarted samba and I was back on the track again...

correct me if I'm wrong, the VM starts killing processes when it has no swap 
left...but I have 280MB of it!
comments are wellcome! (I need more educated opinions on this)

dragos