Re: Version 7.12uc1, killing 4ge's

In article <4pme4b$8d3@nic.global-one.no>, Nils Myklebust <Nils.Myklebus t@ccmail.telemax.no> writes >ktomchik@homedepot.com (Kate Tomchik) wrote: > >>I am using 7.12UC1 on HP 9.04, testing to see if we should upgrade >>our systems from 5.03. During our test, the shared memory dumped >>into the /tmp and filled it up (I know, I can turn this dumping off). >>The 4ge that was causing the problem was killed, then the whole >>server crashed. Now I know that in version 5.03, you always need to >>kill the calling job, never the generated sqlturbo to avoid a crash, >>but version 7 has no sqlturbos! How do I avoid this problem? > >We kill 4gl executables every now and then without any problems >(OnLine 7.10 on SCO 3.2.4.2). It shouldn't cause any problems except >perhaps if the executable is inside a transaction. We haven't tested >that. >Are you shure the crash was due to you killing the 4ge, and not some >other cause. > Did the OnLine log say that killing the 4ge crashed the server. When /tmp fills up OnLine will eventually die and Unix tend to crash as well. The correct way to kill 4ges is to use:- onstat -u to get the session id ( look at the user and tty and compare this will the values from ps -ef / ps -auxg for the running 4ge). tbmode -z <session-id. to kill the session. Cheers, -- David Williams