Hello!
I am using an oracle 10.2.0.4 on HP-UX 11.23
Here is the error message from my log file:
ORA-03113: end-of-file on communication channel
25.05.2008 18:28:56 my_db_package.insert_function ERROR end with exception at "ERROR at update into my_table" at record number ...
25.05.2008 18:28:56 my_db_package.insert_function ERROR unhandled exception ORA-12899: value too large for column "MY_SCHEM
A"."MY_TABLE"."TIMESTAMP" (actual: 42, maximum: 11)
25.05.2008 18:59:46 my_db_package.insert_function ERROR end with exception at "ERROR at update into my_table" at record number ...
25.05.2008 18:59:46 my_db_package.insert_function ERROR unhandled exception ORA-12899: value too large for column "MY_SCHEM
A"."MY_TABLE"."TIMESTAMP" (actual: 28, maximum: 11)
25.05.2008 18:22:37 my_db_package.insert_function ERROR end with exception at "ERROR at update into my_table" at record number ...
25.05.2008 18:22:37 my_db_package.insert_function ERROR unhandled exception ORA-12899: value too large for column "MY_SCHEM
A"."MY_TABLE"."TIMESTAMP" (actual: 25, maximum: 11)
So my questions are:
1. Should I consider the three ORA-12899 errors? Can I trust the output of the program after an ORA-03113 occurs? I mean, if ORA-03113 always leads to unexpected behaviour, then I should only consider it, and pay no attention to all following errors, right?
2. Is it possible that insufficient memory caused the ORA-03113? I tried to increase the SGA (to 12GB!) and then the program completed. However, the some (this time four) TIMESTAMP problems occured.
3. How can I handle the TIMESTAMP problem? The filed TIMESTAMP is set by a before insert trigger. The trigger is defined like this:
create or replace trigger my_table_bi
before insert on my_table for each row
begin
if (some condition ) then
RAISE_APPLICATION_ERROR(-20400,'Conflict: bla bla bla');
end if;
:new.timestamp := systimestamp;
end my_table_bi;
The filed TIMESTAMP is defined in the table MY_TABLE like this:
.....
timestamp TIMESTAMP NOT NULL,
.....
I think these declarations are normal. So does anyone have an idea what could have caused the problems with the TIMESTAMPs?