Conference ulysse::rdb_vms_competition

1270.0. "ORA 62/VMS BUGCHECK?" by UNYEM::UTTERD (Dean W. Utter) Thu Jul 15 1993 21:22

    HAS ANY ONE ENCOUNTERED A PROBLEM USING ORACLE V6.0.36.7 IN A VMS
    V5.5-1 CLUSTER USING THE PARALLEL SERVER CAUSING VMS TO CRASH WITH A
    FATAL BUG CHECK ERROR OF - DOUBLY DEALLOCATED MEMORY BLOCK ?
    
    CALLS TO ORACLE INSIST THEY HAVE NOT HEARD THIS PROBLEM AND INSIST IT
    MUST BE VMS.  OF COURSE, THE ONLY TIME I GET THE BUG CHECK IS WHEN THE
    SECOND NODE IN A 2-NODE VAX CLUSTER HAS MOUNTED AND OPENED THE SAME
    ORACLE DATABASE IN SHARED MODE AS IS MOUNTED SHARED ON THE 1ST NODE.
    
    EVERYTHING IS FINE IF I JUST MOUNT/OPEN SHARED THE PRODUCTION DATABASE
    ON ONLY ONE NODE.  WHEN THE SECOND NODE MOUNTS/OPEN SHARED THE SAME
    PRODUCTION DATABASE THE 1ST NODE CRASHES WITHIN ABOUT AN HOUR OR TWO.
    
    INITIAL LOOK AT CRASH DUMP SHOWS VMS ATTEMPTING TO DEALLOCATE A SYSTEM
    SUBROUTINE-GENERATED TQE AND IS USING A BOGUS ADDRESS TO CHECK FOR THE
    SIZE OF THE TQE TO DEALLOCATE FROM NON-PAGED POOL.
    
    ANY ONE WHO HAS HEARD OF THIS PROBLEM OR CAN GIVE ME SOME INSIGHT IN
    HOW TO INTERPRET THE CRASH DUMP SO THAT I CAN PROVE IT IS ORACLE'S
    FAULT WOULD BE GREATLY APPRECIATED!!!!!
    
    THANKS IN ADVANCE,
    
    DINO :^(