[archive] SP2 and Thin Client

[Migrated content. Thread originally posted on 24 March 2006]

Ping responses are really the same.

I've tryed to modify "Tcp receive Window" value but it doesn't change anything.

Is there other parameters to tune ?

:(

Hi,

We are developping Client/Server Application.
We used thin Client.

The Server is UNIX AIX 5.2. Acurcl runs on it
and clients (Windows Stations) communicate with server thru The Thin Client.

It works very well when clients are WinXP or WinXP SP1.
But when clients ars WinXP SP2, time delays are multiplied by 10 or more.

We've disabled firewalls, antivirus but it's not better.

Is there a parameter or environment variable to set?

Regards

[Migrated content. Thread originally posted on 24 March 2006]

Ping responses are really the same.

I've tryed to modify "Tcp receive Window" value but it doesn't change anything.

Is there other parameters to tune ?

:(

Have you tried to ping and compared the ping responses? Are they the same or different as well?

[Migrated content. Thread originally posted on 24 March 2006]

Ping responses are really the same.

I've tryed to modify "Tcp receive Window" value but it doesn't change anything.

Is there other parameters to tune ?

:(

In general, I've noticed that SP2 slows systems down a lot. I'd be interested if anyone knows of ways to speed it up!

Rob

[Migrated content. Thread originally posted on 24 March 2006]

Ping responses are really the same.

I've tryed to modify "Tcp receive Window" value but it doesn't change anything.

Is there other parameters to tune ?

:(

There is a number of options to try, but I think you should talk with Tech support about it as there is there is not enough information to give an exact answer.

[Migrated content. Thread originally posted on 24 March 2006]

Ping responses are really the same.

I've tryed to modify "Tcp receive Window" value but it doesn't change anything.

Is there other parameters to tune ?

:(

Hi,

We had this problem on a few of our customer sites (we use AIX 5.2 too). It turned out to be a setting on the network card on the server, if the network card was a newer gigabit card.

It's something to do with parameters for large_send and chksum_offload

They will be set to yes as default but need changing to no

Please note all users must be logged off the system as a reboot is required after running the command.

Command;

chdev -l ent0 -a large_send=no -a chksum_offload=no -P
shutdown -Fr

Playing round with the duplex/speed of the pc client cards also had some improvement (i.e. changing to half duplex, and/or dropping speed to 10).

This took us ages to figure out as neither ourselves nor Acucorp could recreate the problem. We're still not 100% sure why this fixes the problem, but it does. This is what worked for us anyway, so you may want to give it a try!

Cheers,

Mark