ICE
TCP/IP Products => TCP-PRO => Topic started by: ruthan on July 27, 2004, 10:48:08 am
-
Inquiry from client reads:
Versione TCP-PRO 5.1.21 32 bit.
We have a problem on a pc XPpro and ICElp
Simetimes it's an error "host doesn't respond", and it breaks.
Can we bypass error's message ?
The network is funtionally.
-
That response means that at the time of a polling interval, the Unix machine is not responding! The solution is to see why this is happening. Some possibilities are that the Unix machine is overloaded and responding too slowly to the connect (check this on the unix machine with the top command or equivalent), the Unix machine is running out of sockets (this is a kernel tunable parameter in SCO) check the socket problem by doing a netstat -a on the Unix machine. Another possibility is that the network is really going down periodically. You didn't say whether it was a lan or wan connection. There are known problems with some DSL equipement and MS VPN's over wans.
The bottom line is to treat the problem, not the symptoms.
-
Client inquires:
Can we modify ICElp's configuration, for example timeout seconds, to resolve the problems?
It's necessary to close/reopen ICElp if this is happening?
-
You can only lengthen the polling interval which may or may not help depending on where the problem is. If the problem is that too many icelp's are connecting at nearly the same time this will help, if it's a case of machine load or running out of sockets it will not help.
The polling interval is changed by right clicking on icelp on the system bar and doing a configure ICELP. The changes will stay in effect after reboot as well.