Re: SE locks-up

Joseph Cullipher wrote: > > community. First we are using SE 7.10 on SCO 3.2v4.2 TCP 1.2.1 on a > Compaq system pro. The sqlhosts file is set for local loopback. The > problem is at times (2 or 3 sometimes more per week) the engine locks-up. > There is no way to be able to connect. We have to get all our users off. > kill the sqlexec process and then restart it. There are no error messages > anywhere. The only 'clue' we have been able to find is that when the > engine locks-up if we do a netstat -a there are some processes in a > FIN_WAIT_2 state that stay there until the sqlexec process is killed. We have a similar problem on a sparc-board running solaris 2.5 and SE 7.20 UC2. We could isolate our problem to the point that when an embedded application does several connects and disconnects to the database the new created sqlexec processes do not die after the disconnect. So our process list grows on and on with Zombies until the embedded application is terminated. At that time all the defunkts disappear. In the moment we use the workaround to leave the connection established all the time. But perhaps anyone can help us to solve the problem generally.