memory usage on Scyld slave nodes

[Daniel Ridge]

Dan,

> I recently installed Scyld Beowulf on our cluster.  It works fine, but when
> I use BeoStatus to look at the slaves it shows that the slaves all have
> about 400MB of their memory taken up.  Anyone know what is eating up the RAM
> on those slaves?  Each node has a total of 512MB of RAM.  Why is beowulf
> eating up so much of it?  Thanks.

Try 'bpsh 0 free' for detailed statistics. Is this memory just buffers?

Otherwise, if you are using a ram disk root, you could accidentally
have specified a ramdisk size which is an order of magnitude too large
in /etc/beowulf/fstab.

Regards,
	Dan Ridge
	Scyld Computing Corporation