Can someone help...? I am a relative newcomer to Unix and need some help with the above. I have inherited from my job predecessor a backup script which backs up files from its 'home' server onto a robotic tape library and then makes a remote connection to another server and does a remote backup of certain selected files. Last week the remote machine went down and having got it back up and running again, the backup script no longer performs the remote connection. All I get are the following error messages: TCP: t_rrcmd: Error setting up stderr: event received on stout. TCP: t_rrcmd: The remote host may have closed the connection. TCP: t-rrcmd: Network error. and... Not on system console. Connection closed. I have studied man pages and TCP/IP manuals and have tried to amend the .rhosts file with little success. I can now resh into the remote machine using my own login but if I try to resh using root I still get a password prompt, which is no good because the backup script need to run automatically. Any help will be GREATLY appreciated!! ps. Sorry if I am missing something obvious Thanks Cameron

I don't know if this will be much help but maybe worth a shot. If you try to telnet to root on a default install of Solaris you will get that not on system console error message. The system will not allow a root remote login by default. You can change that by editing an entry in (I think -- at home so can't check) /etc/default/login file. It will be obvious when you see the file -- just comment out the appropriate entry. I'm skeptical that this is the problem as the file should not change just because the system went down. Please let me know if you figure out the problem. Good Luck, Jon

Thanks Jon! I'm getting there... I can now login as ROOT on the remote server but whenever the script tries to use the following line: resh sas-seb df -k the following error msg appears on the remote machine's terminal vdu: Can't get stderr port; Resource temporarily unavailable and on the 'home' server where the script is running, I get the following: TCP: t_rrcmd: Error setting up stderr: event received on stout. TCP: t_rrcmd: The remote host may have closed the connection. TCP: t_rrcmd: Network error. Do I need to do something on the remote machine, and yes it is running Solaris 7. Many Thanks Cameron

Hi Jon, I think I know what the problem is. After checking other postings, I ran a 'netstat lsof' command on the remote system for open TCP sockets. WOAH!! These ran into THOUSANDS! The majority of them were in a state of TIME_WAIT, so I ran a search on TIME_WAIT on Google.com (v v useful tool) and found a set of notes that said if there are a lot of sockets in TIME_WAIT then connections to that machine will be rejected! And there was a lot of connections from the Host to the Remote machine in this state. What I now need to know is how do I get rid of all these TIME_WAIT connections? Short of rebooting the remote machine, which I really don't want to do! Hoping you can help... Thanks Cameron

Hi Jon, After I ran a netstat query, I noticed that the majority of the ports that were on TIME_WAIT were all the same number, ie for JAVA. This was interesting as I don't run JAVA JDK on that server. So I rebooted the server and was able to resh in from root no problems :o) I think the build up of JAVA TIME_WAITS was due to the fact was that this server runs SAS and is used heavily by users who connect through a JAVA interface from there desktop PCs. When the server went down, it was a busy time of day and there were a lot of users logged on. I think this is where the JAVA connections have came from and have not closed themselves down. So a fresh reboot of the server solved this problem. So if this happens again, at least I'll know what to do.... Many thanks for all your help! Cameron

You modify your TCP values using the ndd command. You should limit the amount of tcp_time_wait parameter for Solaris 8 its: ndd -set /dev/tcp tcp_time_wait_interval 60000 that lowers it from the standard 240000 .. which is a little better.