Oracle Internals

If you ever upgrade your database running on a RAC cluster (including Exadata) to version 19c and then rollback your operation back to previous version and you encountered error " ORA-29702: error occurred in Cluster Group Service operation" while starting your database. Additionally you can start the database with another name but not the same database which has been upgraded and now being rollback.  This is due to  Bug 31561819 and it's related to specifically with RAC cluster. Mentioned bug exists from version 12.1.0.2 to 19.8.0 and later included in patch set itself. So if you need to fix this problem you would need to apply this patch. As a workaround none is mentioned in oracle support but there is one: Workaround : Perform the cold restart services of the your cluster. crsctl stop crs (on all nodes sequentially) crsctl start crs (on all nodes sequentially) Once restart of cluster services is done, try starting the database back and it should work as in our case. If

There would be occasions where you would upgrade your oracle database which is hosting your APEX and post upgrade you get error while opening APEX URL " Error Processing Request Contact your application administrator " This looks very strange as you have just upgraded only database. Database version and APEX version are compatible. For Argument lets say both are on 19c.  Now since UI doesn't give detailed information about the error you would need to get to backend oracle database and query view apex_debug_messages.     select * from apex_debug_messages; Looking at the results of this view you would get the clue of the error in this case it was:      ORA-04063 package body APEX_190100.WWV_FLOW_SESSION_RAS has error This means this package body has problem and looking at the status of this body it's found to be invalid and when you try to compile it get following error      alter package APEX_050100.WWV_FLOW_SESSION_RAS compile body;      Warning: Package Body altered