Hi Rich,

Its not your GS116 switch. I'm using a GS116 as my home switch ,
 and I dont have  this problem with my switch.

 I have a  Sunray 1G   and a SunRay 2   for testing  purposes  at home
and they are  directly  attached  to this  switch.

I have seen  bad graphics  performance  with two  scenarios :

   1)   A  switch port  is  set to  fixed  speed and duplex.
This makes the sunray belive its attached to a Stupid HUB and the sunray goes into 100 Mbps half duplex.

    2)  A Switch with  1Gbit  uplink  and  100  Mbit  ports . The
Switch typically handles TCP packets reliably but does not buffer UDP packets , the UDP packets gets discarded and need to be retransmitted from the servers.



In your case it would be interesting to see the server statistics on the NIC serving th sunrays,

# netstat -i should show if you have tx-errors on the interface where the Sun Rays are connectd.

SUNSOLVE system handbook says that the initial X2100 server moodel has two I/F:es

Two 10/100/1000Base-T Ethernet ports (nge0=LAN1 and bge0=LAN2)

I would prefer to use the Broadcomm port ( LAN2 ) for sunray connections . the NVIDIA Ethernet port on the nforce 4 chipset is not really a professional
quality  gigabit  ethernet device.  Its a  Home PC network chip.


Next  thing to check is of course   the  Network  cable  from the  server ,
make sure its CAT 6.


//Lars



Rich Teer skrev:
Hi all,

This seems to be a recurring theme here, but here goes...  I've
installed a very small Sun Ray system where my wife works, and
they're complaining about poor graphics performance, especially
when scrolling a browser window.  The effect is reminiscent of
the bad old days of interlaced images over a dialup connection.
(Applications seem to run at an acceptable speed, just screen
refreshes could be improved.)

I've a few things here which might be applicable, but I'd like to
get to the bottom of this issue and solve it once and for all!

Here's some output from utcapture:

# TERMINALID      TIMESTAMP TOTAL PACKET   TOTAL LOSS   BYTES SENT PERCENT LOSS 
     LATENCY
00144fe4c5b9 20081204072607 159199432 26549517 1228284694 21.566 00144fe4c5b9 20081204072637 159199666 26549575 1228423758 24.786 00144fe4c5b9 20081204072707 159204536 26550855 1232240548 26.622 00144fe4c5b9 20081204072737 159209024 26552099 1235935770 27.718 00144fe4c5b9 20081204072752 159223822 26558363 1244974054 42.330 00144fe4c5b9 20081204072807 159232746 26561859 1251139582 39.175 00144fe4c5b9 20081204072837 159261524 26575784 1266804516 48.388 00144fe4c5b9 20081204072852 159293508 26591384 1284242436 48.774 00144fe4c5b9 20081204072907 159316291 26601802 1297510370 45.727
If I'm reading that correctly, that's between 20 and 50% packet loss,
which would explain a lot!

The server is a Sun Fire X2100 with a 2.2 GHz Opteron 148 CPU,
4 GB of RAM, and dual 80 GB disks.  They are using a 16-port GBE
switch, specifically a Netgear GS116 with CAT 5e cabling.  There
are only two thin clients on this network: a Sun Ray 2 FS and a
Sun Ray 2; both have 22" inch monitors.

The operating system is build 82 of Nevada (a little old, I know),
and they're using SRSS 4.0 with patch 127554-03 applied.  The
thin clients have been power cycled and are running the updated
firmware contained in that patch.

I briefly tried swapping out the GBE switch for a Netgear 100 baseT
one, but that didn't seem to make much (if any) difference.  Any
help greatfully received!

TIA,

_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users

Reply via email to