1 2 Previous Next 15 Replies Latest reply on Aug 13, 2019 2:34 PM by pwiseman Go to original post
      • 15. Re: Issue with system() call with libclntsh.so.19.1
        pwiseman

        Thank you cj for logging the bug.  We've been able to reference that in a couple of SR's we have open on this issue (slightly different symptoms, but we believe the same root cause), but as yet, despite an Escalation, despite being accepted as a Sev 1, despite being reproduced, despite being passed through to development, there has been no progress, no recognition from development that there's a serious issue here, no information on what the thread is being used for.  Apparently there is a fix in 20.1, but that doesn't help us when 19 is the long term support release.  We have been provided with no information on how it's been fixed - has the thread been removed, is there now a mechanism to disable it?  As user11763611 has pointed out - it's not something that a customer can solve without having further information from Oracle on how to disable the thread, or disable signals being caught by the thread.  We have set DIAG_SIGHANDLER_ENABLED=FALSE but that isn't helping.  We end up with multiple scenarious: On Linux we're getting random coredumps when the signal is caught by the oracle thread.  On AIX, the signal is lost, and this is the method by which processes are terminated, hence we can't shut down cleanly.  (SR 3-20270713651, SR 3-20695062651)

        1 2 Previous Next