Strange.  I've run 1.4 with client + MDS + 5 OSTs in a single 96MB UML.
The memory allocated by lustre generally scales as a function of RAM in
the system, so that it can at least function in lower-memory environments.
That said, I haven't run 1.6 in this environment.

Well since my experience in xen I know he's using an unstable version
of xen to work from since xen 3.0.4 is a 2.6.16.33 based system and
only their scm repos have the 2.6.18. Also I've seen funny things
happen when a dom0/U starts running out of memory with xen. It really
depends on how your system is setup, I've seen xen kill processes
simply because its out of memory.

Also if you give dom0 512M of memory you really end up with 437M shown
by /proc/meminfo so depending on how small he's got memory set to it
might have less than he thought. I think this has something todo with
the xen.gz hypervisor taking up space.

_______________________________________________
Lustre-discuss mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-discuss

Reply via email to