WSNAT_CAT:1023 Error while connecting to Workstation Clients

Question

Hi All,
I am trying to run around 12000 Workstation Clients, and want to connect them all to a single node. But I am getting the Following errors over and over again:
1.) WSNAT_CAT:1023: WARN: No client context space available in handler
2.) WSNAT_CAT:1175: WARN: Disconnecting listener connection to workstation, request timed out
3.) TDM_CAT:3000: INFO: Client connection request timed out
Following are the snippets of my ubbconfig file :
MAXACCESSERS 21000
MAXSERVERS 5000
MAXWSCLIENTS=15000
I have set "TCPKEEPALIVE=Y" in dmconfig environment, in place of using "-K" option in CLOPT. Is this correct?
Following is the CLOPT for my WSL Server (I have 70 such WSL processes running) :
CLOPT=" -- -n //xx.xx.xx.xx:23040 -m 10 -M 10 -x 20 -p 40000 -P 40199"
I would like to explain this. I have MAXACCESSERS = 21000, MAXWSCLIENTS=15000 and MAXSERVERS=5000. So this cannot be an issue. To support my 12000 workstation clients, I have given:
-m=10 (Minimum 10 WSH processes)
-M=10 (Maximum 10 WSH processes)
-x=20 (Multiplexing factor=20, i.e. one WSH will connect(multiplex) with 20 Workstation Clients (WSC)).
According to this configuration, I should be able to support : 70 (WSL) * 10 (WSH) * 20 (WSC) = 14000 Workstation Clients. I am not able to understand then why only to start 12000 Workstation clients, Tuxedo is giving the above errors??
Do I need to change any configuration settings? Urgent help will bw of utmost value.
-Thanks and Regards
-Tuxedo Developer

Answer 1

Hi Steven,
Yes, you are correct to observe that one WSL will serve upto 200 Workstation Clients. I shall jot the information pointwise :
1.) 70 Such WSL's are booted, with each capable of serving upto 200 WS clients (10*20).
2.) You are right. All the 12000 Workstation clients are started using a script. But I had known the concern of all requesting the data at the same time, so each of the clients were started with a delay over the previous client.
3.) The clients are built in such a way, that they request a service every 2 minutes, and are all alive all the time.
4.) As per the load balancing is concerned for the WSL's, I have put a script which changes the environment variables (WSNADDR) in such a way, that after every second, a new WSL's address is issued for the client to connect to. As I have 70 WSL processes up, the script sets the WNSADDR of each WSL for one second only. So all the WSL's are naturally load balanced.
So with this configuration, I should be able to successfully run upto 14000 clients, but I am having problem with even starting 12000 at a time.
Also I am amazed at the error WSNAT_CAT:1023, which in manual describes a rare condition. It suggests that there is no more context space in the handler. I am not sure how this is possible.
I also forgot to mention that, I also had to force shutdown some of the clients, who emitted the WSNAT_CAT:1287 Warning in the ULOG, but I think that should not be an issue. Do you think that killing clients abruptly may cause weird behaviour? Or can they for instance corrupt any WSH, or make the sockets go crazy?
-Thanks in Advance
-Tuxedo Developer