Hi, this is my first post to the list. I think I've got this problem since last tuesday when I've put in production state my first SLES9. I have lots of connections in TIME_WAIT state. As I understand the TIME_WAIT state is used to indicate that a closing socket still have to manage some packets in the network before "disapear". My /proc/sys/net/ipv4/tcp_fin_timeout is set to 30 seconds, and apparently most (all?) connections are ending by this timeout, not by themselves. Most of the sockets interconnect processes running in the very same machine, I don't understand why is the timeout reached. I supose this would be a normal situation with a slow machine, but this is a new one with 2 Xeon processors @ 3GHz and 2 GB of ram, it only have to handle 1 Mbps of internet connection. I cannot put into production the old machine to compare them. As as second problem the machine stales for various seconds (5 to 20) from time to time. I've also changed the NIC, but the situation is still present. I've applied all possible patches, the OS is up to date. Running processes: apache2 (prefork) postfix imss, interscan antivirus daemon, postfix content filter netstat -nep see the attached file. Best regards, Ernesto Silva.
participants (1)
-
Silva, Ernesto - Webmaster