[K12OSN] OT: Testing FC10 results

David Hopkins dahopkins429 at gmail.com
Sat Apr 4 22:05:46 UTC 2009

>> So, could it be that the ltspbr0 bridge is a source of the issue?
> I don't think that should impact performance much at all. IIRC the chroot
> memory footprint in K12Linux is still higher than K12LTSP. That might be
> part of it, if your clients have less that 128MB of memory.

All clients have at least 128Mb, most have 256Mb.

>> Or
>> using nbdswap for the clients?
> Actually, NBD swap is supposed to be faster than NFS swap, that's why the
> developers implemented it.
>> Any ideas on how to tell?
> I thought it might be the LDM_DIRECTX setting, but that is set to TRUE by
> default on Fedora, which means that X traffic isn't encrypted on K12LTSP by
> default. So it must be something else.  Higher network traffic for some
> reason, like increased screen resolution or color depth? You said the server
> has plenty of memory and CPU, and I don't think the disk IO should change
> that much.

In terms of bandwidth, this server is performing the same for disk I/O
as the other servers running 4.2 (measured using netio).  Screen
resolution is likewise typical.  I could force the color depth to 16
instead of 24. It is confusing.  I am using ldap for authentication
and wonder if something changed wrt to how authentication is working,
or if dns has changed?  The host table is empty for ltsp5?  ltsp4.2
had an entry for every client.  This is the one difference that I know
of since the clients are using 172.x.y.z whereas 4.2 used 192.x.y.z.
I can change this but don't really see why it would have an effect.

Dave Hopkins

More information about the K12OSN mailing list