I wrote a pipelined function to query data from a remote database. I keep getting
ORA-06502: PL/SQL: numeric or value error: character string buffer too
small.
I think I do understand when this error would occur, for example when a table column is defined as VARCHAR2(10) and you try to insert something bigger than 10 byte. But in this case, I really don't see whats wrong.
Perhaps first I show the parameters of the local and the remote database. I think it might be important, that on both DBs NLS_LENGTH_SEMANTICS is set to BYTE.
Local DB (where the pipelined function is stored):
params local db
Remote DB (from where data is queried):
params remote db
Now the code in local db:
create or replace function f_get_pl_data return tb_pl_palette pipelined is
begin
for i in (select p.*, 123 LAGER from palette#myremotedb p)
loop
pipe row(tt_pl_palette(i.pid,i.bereich,i.regal,i.fach,i.ebene,i.vol_klasse,i.lhm_typ,i.zustand,i.neu_datum,
i.neu_zeit,i.neu_usr,i.aender_datum,i.aender_zeit,i.aender_usr,i.verl_datum,i.tournr,
i.fil_nr,i.retournr,i.fz_nr,i.fahrer_nr,i.eroeff_auswahl,i.tpa_knz,i.lfsaender_knz,
i.verladen_am,i.verladen_um,i.verladen_von,i.verladen_von2,i.leer_gew,i.soll_gew,
i.ist_gew,i.lager));
end loop;
return;
end;
CREATE OR REPLACE TYPE "TT_PL_PALETTE" AS OBJECT (
pid VARCHAR2(14),
bereich VARCHAR2(2),
regal VARCHAR2(2),
fach VARCHAR2(3),
ebene VARCHAR2(2),
vol_klasse INTEGER,
lhm_typ INTEGER,
zustand INTEGER,
neu_datum DATE,
neu_zeit VARCHAR2(11),
neu_usr VARCHAR2(4),
aender_datum DATE,
aender_zeit VARCHAR2(11),
aender_usr VARCHAR2(4),
verl_datum DATE,
tournr VARCHAR2(6),
fil_nr VARCHAR2(4),
retournr VARCHAR2(10),
fz_nr INTEGER,
fahrer_nr INTEGER,
eroeff_auswahl INTEGER,
tpa_knz VARCHAR2(1),
lfsaender_knz VARCHAR2(1),
verladen_am DATE,
verladen_um VARCHAR2(11),
verladen_von VARCHAR2(4),
verladen_von2 VARCHAR2(4),
leer_gew NUMBER(7,3),
soll_gew NUMBER(7,3),
ist_gew NUMBER(7,3),
lager NUMBER
)
CREATE OR REPLACE TYPE "TB_PL_PALETTE" as TABLE OF TT_PL_PALETTE
And this is the table specifications on remote db:
create table PALETTE
(
pid VARCHAR2(14) not null,
bereich VARCHAR2(2) not null,
regal VARCHAR2(2) not null,
fach VARCHAR2(3) not null,
ebene VARCHAR2(2) not null,
vol_klasse INTEGER,
lhm_typ INTEGER,
zustand INTEGER,
neu_datum DATE default trunc(sysdate),
neu_zeit VARCHAR2(11) default to_char(sysdate, 'HH24:MI:SS'),
neu_usr VARCHAR2(4),
aender_datum DATE,
aender_zeit VARCHAR2(11),
aender_usr VARCHAR2(4),
verl_datum DATE,
tournr VARCHAR2(6),
fil_nr VARCHAR2(4),
retournr VARCHAR2(10),
fz_nr INTEGER,
fahrer_nr INTEGER,
eroeff_auswahl INTEGER,
tpa_knz VARCHAR2(1) default '0',
lfsaender_knz VARCHAR2(1) default 'N',
verladen_am DATE,
verladen_um VARCHAR2(11),
verladen_von VARCHAR2(4),
verladen_von2 VARCHAR2(4),
leer_gew NUMBER(7,3),
soll_gew NUMBER(7,3),
ist_gew NUMBER(7,3)
)
So you can see, when I created the type in local db I choosed the exact sizes of the columns in remote db.
But when I execute/query the pipelined function, I get this error (sorry it's german, but I wrote english in title):
error message
How can this happen? Do you have any idea what's wrong?
Appreciate any help, thanks!
EDIT 2021-03-02:
#ShaunPeterson Thanks for reply, NLS_CHARACTERSET is set to AL32UTF8 on both DBs, local and remote.
I just found out, the problem seems to be caused by 2 different IDEs that are used in our company.
I am using PL/SQL Developer from allround automations. When I posted the table specifications on remote DB (table "PALETTE"), I connected to remote DB using that IDE, and it showed the column types/sizes as you can see in my original post. I repeat the first 6 columns:
create table PALETTE
(
pid VARCHAR2(14) not null,
bereich VARCHAR2(2) not null,
regal VARCHAR2(2) not null,
fach VARCHAR2(3) not null,
ebene VARCHAR2(2) not null,
vol_klasse INTEGER,
But when Oracle SQL Developer is used, then it looks like this:
CREATE TABLE "PSTEDI"."PALETTE"
( "PID" VARCHAR2(14 CHAR) NOT NULL ENABLE,
"BEREICH" VARCHAR2(2 CHAR) NOT NULL ENABLE,
"REGAL" VARCHAR2(2 CHAR) NOT NULL ENABLE,
"FACH" VARCHAR2(3 CHAR) NOT NULL ENABLE,
"EBENE" VARCHAR2(2 CHAR) NOT NULL ENABLE,
"VOL_KLASSE" NUMBER(*,0),
So it seems PL/SQL Developer just shows wrong specifications. I have no idea why it is like that, and I would like to know, but that's another question.
I solved my problem using the column types and sizes Oracle SQL Developer shows.
Thank you.
#Dietz, what you said was fine for me.
I'm having the same problem as you, I am wondering if it's a bug. Like you:
I created types in ADT exactly as in their respective table columns.
I set my NLS_LENGTH_SEMANTICS.
However:
I set a default in SQL Developer.
I am not accessing a remote DB.
I don't believe it's related to either of these.
FYI, I'm going to try one more thing and if it doesn't work, I'll open a ticket with Oracle Support. I will update here with what I find or workaround from Oracle Support.
I have a table with several columns, like this:
CREATE TABLE CRM.INFO_ADICIONAL
(
ID_INFO_ADICIONAL NUMBER(10) NOT NULL,
NOMBRE VARCHAR2(100 BYTE) NOT NULL,
OBLIGATORIO NUMBER(1) NOT NULL,
TIPO_DATO VARCHAR2(2 BYTE) NOT NULL,
ACTIVO NUMBER(1) NOT NULL,
ID_TIPO_REQUERIMIENTO NUMBER(10) NOT NULL,
ID_USUARIO_AUDIT NUMBER(10) NOT NULL,
ORDEN NUMBER(3) DEFAULT 1,
RECHAZO_POR_NO NUMBER(1),
ID_TIPO_ARCHIVO_ADJUNTO NUMBER(10),
SOLICITAR_EN VARCHAR2(30 BYTE),
ID_CONSULTA NUMBER(10),
COMBO_ID VARCHAR2(40 BYTE),
APLICAR_COMO_VENC NUMBER(1),
MODIFICABLE NUMBER(1) DEFAULT 0,
ID_AREA_GESTION NUMBER(10),
ID_TAREA NUMBER(10)
)
The "COMBO_ID" column is the target. It is defined as VARCHAR, but when I'm trying to insert a row, TOAD displays
"ORA-06502: PL/SQL: error : error de conversión de carácter a número
numérico o de valor"
Or a 'numeric conversion error', in english.
This table have some pre-existing data, and I even found some rows including values at COMBO_ID column, all of them being VARCHAR, i.e.:
NACION (Nation), SEXO (Sex), etc
I tried a few simple SELECT statements
SELECT
ID_INFO_ADICIONAL,
NOMBRE,
OBLIGATORIO,
TIPO_DATO,
ACTIVO,
ID_TIPO_REQUERIMIENTO,
ID_USUARIO_AUDIT,
ORDEN,
RECHAZO_POR_NO,
ID_TIPO_ARCHIVO_ADJUNTO,
SOLICITAR_EN,
COMBO_ID,
APLICAR_COMO_VENC,
ID_CONSULTA,
MODIFICABLE,
ID_AREA_GESTION,
ID_TAREA
INTO
pRegistro
FROM
crm.info_adicional
where pRegistro is declared as
pRegistro INFO_ADICIONAL%ROWTYPE;
Again, I'm still getting this 'numeric conversion error'.
But, wait, if I hardcode the SELECT value in COMBO_ID column with a NUMBER:
SELECT
--other columns
123456 COMBO_ID,
--other columns
INTO
pRegistro
FROM
crm.info_adicional
It works, what the heck, it's defined as VARCHAR.
If I do the same but harcoding a string, it fails to execute again
Already tried in my DEV environment, and it's working fine.
I'm not a pro in Oracle, but I feel pretty lost.
Could it be that tables get "confused"?
Any clues?
That error can also be raised if you try to push a character string that is longer than your VARCHAR2's capacity (40 in your case).
Try to check if all the data you are trying to insert is correct :
SELECT
COMBO_ID
FROM
crm.info_adicional
ORDER BY length(COMBO_ID) desc;
That would also explain why it works fine on your DEV environment which, I suppose, has different data.
Okay, I already found the answer.
Quoting Oracle Documentation:
The %ROWTYPE attribute provides a record type that represents a row in a table or view. Columns in a row and corresponding fields in a record have the same names and datatypes.
So, basically, the SELECT statement needed to be in the same order as the table columns definition.
In my case, I had a few columns (including COMBO_ID) in a different order.
Tried, re-ordering, and works like a charm.
Thank you all for the support.
I have a weird issue. In one of the production DB servers we are trying to run
select * from SCHEMA_NAME.TABLE_NAME
When we run the query we are getting invalid identifier error. But when we select a particular column and run the same select query on the same table , we are able to get the output.
Please help me to understand the root cause.
Type created:
create or replace
TYPE "TEST_TYPE" is object
(
MULTIROW_ID VARCHAR2(100 CHAR),
LOSS_ENTRY_TYPE VARCHAR2(1000 CHAR),
SUB_CATEGORY VARCHAR2(1000 CHAR),
LOSS_AMOUNT NUMBER,
LOSS_ENTRY_CURR VARCHAR2(100 CHAR)
);
Type 2 created:
create or replace
TYPE "TEST_TYPE1" AS TABLE OF TEST_TYPE;
Main create table query using above table type columns:
CREATE TABLE "MS_TEST_DATA"
( "REGION" VARCHAR2(4000 CHAR),
"Entries" "TEST_TYPE1" ,
"ILE_DET_FUNCTION_OF_DISC_COM" VARCHAR2(4000 CHAR)
)
NESTED TABLE "Entries" STORE AS "TEST_TYPE3";
If I run this locally, I am able to access the table with select * query.
but in server I don't have all the accesses. There I have read-only access to tables. I can only run select * queries.
Avoid using double quotes in DDL scripts. If it is system generated, write a script to eliminate double quotes wherever possible.
I'm loading some data into oracle using an external table. One of the fields "CREATEDON" is a date but is formatted like this 20151231
How can I convert CREATEDON to a proper oracle date datatype. This is my attempt so far which doesn't work...
CREATE TABLE "AMS"."DATA"
( "BLANK" VARCHAR2(255 BYTE),
"BLANK2" VARCHAR2(255 BYTE),
"CLIENT" VARCHAR2(255 BYTE),
"MATERIAL" DATE,
"CREATEDON" VARCHAR2(255 BYTE)
)
ORGANIZATION EXTERNAL
( TYPE ORACLE_LOADER
DEFAULT DIRECTORY "DIR"
ACCESS PARAMETERS
( records delimited BY '\r\n'
skip 1
fields terminated BY '|' lrtrim missing field VALUES are NULL
)
LOCATION ( "DIR":'foo.txt' )
)
REJECT LIMIT UNLIMITED ;
You have two options:
Keep as is, and transform in your select statements, such as:
select ..., to_date(CREATEDON,'YYYYMMDD') CREATEDON
Define the format in the table DDL:
"CREATEDON" VARCHAR2(255 BYTE) DATE_FORMAT DATE MASK "YYYYMMDD"
I am using ODP.NET with EF in an MVC4 application. When performing the below operation,
_db.Set<User>().FirstOrDefault(x => x.Email == Email || x.DisplayName == DisplayName)
it throws exception
[Specified cast is not valid.] Oracle.DataAccess.Client.OracleDataReader.GetDecimal(Int32 i)
What I don't understand is that my User Object has no Int fields, I made them all decimal to smooth things over with Oracle DB.
My user table:
CREATE TABLE USERS (
USER_ID NUMBER(38) NOT NULL,
DISPLAY_NAME VARCHAR2(50 CHAR),
IS_LOCKED_OUT_FL VARCHAR2(1 CHAR),
IS_LOCKED_REASON VARCHAR2(256 CHAR),
IS_DELETED_FL VARCHAR2(1 CHAR),
EMAIL VARCHAR2(254 CHAR),
TITLE VARCHAR2(20 CHAR),
FIRSTNAME VARCHAR2(150 CHAR),
SURNAME VARCHAR2(150 CHAR),
DATECREATED DATE DEFAULT SYSDATE,
SCHEME_ID NUMBER(38),
STATUS VARCHAR2(20 CHAR) DEFAULT 'New',
PHONE VARCHAR2(20 CHAR) DEFAULT '',
PRODUCT_ID NUMBER(38) DEFAULT 0,
CONSTRAINT BCARD_USERS_PK PRIMARY KEY (USER_ID),
CONSTRAINT UK_BCARD_USERS UNIQUE (SCHEME_ID))
TABLESPACE SYSTEM
STORAGE (
INITIAL 64K
NEXT 1M
MAXEXTENTS UNLIMITED
)
I don't even know where to start debugging because the error is caused by ODP.NET and I can't step into the library. I should also mention that this doesn't happen on my 32-bit development machine, but the same code throws the exception on a 64-bit production machine
using the same 32-bit library. My project was built the target AnyCPU.
To be clear, the Int32 parameter of the GetDecimal method is the index of the column - it's throwing an error trying to convert column with index 'i'. It's not trying to convert anything to an int.
That said, my guess would be that your User object has a decimal field that's not nullable (decimal?), yet some of your columns like scheme_id and product_id allow nulls in the database.