Client_Read_Timeout


Used for:	RTserver only
Type:	Real Number
Default Value:	`0.0`
Valid Values:	Any real number 0.0 or greater

The Client_Read_Timeout option specifies how often (in seconds) data is expected to be available for reading on a connection between an RTserver and an RTclient. This timeout is used to check for possible network failures or ghost clients. If a read timeout occurs, the RTserver sends a keep alive message to the RTclient. The RTserver waits for the number of seconds you specify in Client_Keep_Alive_Timeout to hear back from the RTclient. If the RTserver does not hear back from the RTclient in that time, the RTserver destroys the connection to that RTclient. Checking for read timeouts is disabled if Client_Read_Timeout is set to 0.0. The larger the value for Client_Read_Timeout, the longer the RTserver waits to detect a possible RTclient failure.

If you enable the server-to-client keep alives by setting Client_Keep_Alive_Timeout and Client_Read_Timeout to values other than 0.0, the RTclients connected to this RTserver must be able to process the keep alive messages from the RTserver, using either TipcSrvMsgProcess or TipcSrvMainLoop.

Helpful Read Timeout Tips

If you are enabling keep alives with a read timeout, you can reduce network traffic by setting the options Client_Read_Timeout and Server_Read_Timeout to different values. If they are set to the same value, other than 0.0, the RTclient and RTserver send a keep-alive message at the same time to each other. This is unnecessary, as only one keep-alive message and response is needed to keep the connection open.

The RT process you set to have the smaller read timeout, whether it is the RTclient or the RTserver, becomes the process that generally sends the keep-alive messages. For more efficient RTserver processing, we recommend that you set the value for Server_Read_Timeout to be smaller than the value for Client_Read_Timeout, or set Client_Read_Timeout to 0.0 to disable server-to-client keep alives. This places the burden of sending keep-alives on the RTclient, which only has one RTserver to check on, instead of on the RTserver, which might support many RTclients.