Select For Update Behaves Differently In 9i And 10g
Jul 16, 2007
We have this statement in our database:
select <primary key> from <table> where <primary key> = 1 for update.
Issuing the above statement in 9i, results in a ROW-S (SS) shared type of lock on the table. However, in an identical database in 10g, it results in ROW-X (SX)type of lock. we see all that by qeurying view DBA_DML_LOCKS. Why the different behavior in 10g? Or is it an ORA.INI setting that causes this behavior?
In addition, in 10g the select for update blocks other users trying to access different rows in the table thru oracle forms 6i application. We see this because DBA_DML_LOCKS reports "Blocking" in column "BLOCKING_OTHERS" for the specific row.
I encountering while giving ORDER BY CLASS at report query.
View was created from Master and Detail table with simple join after this i created group above report with order by ID (Varchar2 , Values like 0010101001) but in report it order ruffly( not ascending or descending but randomly) .
I wrote a simple procedure to copy the create timestamp and create user name to update timestamp and update user name of the same record. (See code below)
This works fine for a hard-coded table and primary key column. However, I cannot figure out how to get this to work with dynamic sql.
All my other procs, which don't use SELECT UPDATE FOR work fine with dynamic sql.
Proc that works:
CODECREATE OR REPLACE PROCEDURE proc_set_upd_columns IS CURSOR c1 IS SELECT * FROM mytable FOR UPDATE ORDER BY mycolumn; c1rec c1%ROWTYPE; [code].........
Partial proc that does NOT work:
CODECREATE OR REPLACE PROCEDURE LDEVORE.proc_set_upd_columns ( p_input_table_name IN VARCHAR2, p_pk_id_col_name IN VARCHAR2) IS v_qry_str VARCHAR2(1000); v_cursor_str VARCHAR2(1000); v_create_tmstmp TIMESTAMP; v_create_user_name VARCHAR2(30); [code].......
I'm looking for a solution to select the first row that is not currently locked in a table and insert a record to another table that reference that first row. this is my scenario:
create table ticket ( id number(10) not null,, ticket_type number(1) not null,, is_sold number(1) not null, CONSTRAINT ticket_pk PRIMARY KEY (id) );
cust_id cust_name ticket_id ------------ -------------------- ---------- 1 John 10000004 2 Sara 10000005
my goal is finding the first free ticket ( not sold ) in the ticket table and insert buyer information of that ticket in customer_ticket table. at last I will mark that ticket as a sold one in ticket table with update.
Problem is that the first transaction locks the the first row in ticket table and the second transaction running the same query goes to wait untill the first transaction commit or rollback. However when first transaction finish successfully, second transaction select duplicate id from ticket table that was selected by the first transaction!
I tried to solve problem with "skip locked" and "nowait" options with select for update, but they didn't work.
I'm having a strange case of ORA-01002 Fetch out of sequence when I perform a process in the application.
Application logs, trace files (produced with dbms_monitor), A system trigger to capture the error - all proved that the culprit is a single SQL, *Not a PL/SQL block* That caused the issue:
Select COL1, COL2, COL3, COL4, rowid FROM TAB1 WHERE COL1 = :1 AND COL2 = :2 AND COL3 = :3 ORDER BY COL1 ASC, COL2 ASC, COL4 ASC FOR UPDATE NOWAIT;
binds are: 'AAA' , 10000 , 0
Also, My trigger looks like this:
CREATE OR REPLACE TRIGGER after_error AFTER SERVERERROR ON DATABASE DECLARE pragma autonomous_transaction;
[code]...
I've read some about this error and everywhere it says that it has to do with fetching from an invalid cursor,And all the examples I've observed are of PL/SQL block - never seen any example/explanation of how it happens in a SQL query.
if the OCI maybe is somehow doing things different than the logs/triggers show?
My code executes a Select For Update before updating a table. In some cases the network is disconnected and it causes the lock to hang. Then, I must kill the session in order to realese this lock.
I want to do it automatically. I would like to create a job that kills session that has a lock due to Select For Update that is not alive.
i have two tables test1 and test2. i want to update the column(DEPT_DSCR) of both the tables TEST1 and TEST2 using select for update and current of...using cursor.
I have a code written as follows :
DECLARE v_mydept1 TEST1.DEPT_CD%TYPE; v_mydept2 TEST2.DEPT_CD%TYPE; CURSOR C1 IS SELECT TEST1.DEPT_CD,TEST2.DEPT_CD FROM TEST1,TEST2 WHERE TEST1.DEPT_CD = TEST2.DEPT_CD AND TEST1.DEPT_CD = 'AA' FOR UPDATE OF TEST1.DEPT_DSCR,TEST2.DEPT_DSCR; [code].......
The above code when run says that it runs successfully. But it does not updates the desired columns[DEPT_DSCR].
It only works when we want to update single or multiple columns of same table...i.e. by providing these columns after "FOR UPDATE OF" I am not sure what is the exact problem when we want to update multiple columns of different tables.
I have a base table (Table A) block with multiple records displayed. I need to track audits to this underlying table in the following way:
If user updates a field in the block I want the pre-changed record's audit fields to be set and I need to create a copy of the record with the changed values. Basically any changes will result in the record being logically deleted, and a copy record created with the newly changed values.
Tried to implement in the block's pre-update trigger which will call a package to directly update Table A then Insert into Table A, then requery the block. Is there a clean and efficient way to do this?
UPDATE t_tt_hours a SET a.sak_request = ( SELECT b.sak_request FROM t_requests b, co c
[Code]...
The problem I am having is that it is updating all rows even when it is pulling back a null value for b.sak_request. I've tried adding b.sak_request is not null to the select statement like this:
UPDATE t_tt_hours a SET a.sak_request = ( SELECT b.sak_request FROM t_requests b, co c WHERE b.nam_eds_tracking_id = c.id_dir_track_eds
[Code]...
but it doesn't seem to make a difference. The reason I need to do this is that the difference between where it matches with a valid (non-null) value is 396 rows vs. 12,484 rows which is too time consuming to run on my page.
i want to create a trigger that will update a table when there is an insert or update.i can't across this error that i don't even know what it means "table %s.%s is mutating, trigger/function may not see it".
*Cause: A trigger (or a user defined plsql function that is referenced in this statement) attempted to look at (or modify) a table that was in the middle of being modified by the statement which fired it.
*Action: Rewrite the trigger (or function) so it does not read that table.
CREATE OR REPLACE TRIGGER set_date_end BEFORE INSERT OR UPDATE OF issued ON shares_amount FOR EACH ROW DECLARE BEGIN INSERT INTO shares_amount(date_end) VALUES(SYSDATE); END set_date_end; /
i am reading the columns value from different table but i want to update it with single update statement. such as how to update multiple columns (50 columns) of table with single update statement .. is there any sql statement available i know it how to do with pl/sql.
insert into test_compound values ('user1','1',systimestamp); insert into test_compound values ('user2','2',systimestamp-4); insert into test_compound values ('user3','3',systimestamp-6);
CREATE OR REPLACE TRIGGER trigger_test FOR UPDATE ON test_compound COMPOUND TRIGGER TYPE t_tab IS TABLE OF VARCHAR2(50); l_tab t_tab := t_tab(); [code].......
When I execute :
update test_compound set last_updated_on=systimestamp where userid='user1' and app='1';
The trigger should update the first row and all the data from test_compound table where userid='user1'. Maybe the problem is that updating the same table inside the trigger is firing in a recursive way the trigger.
I have a column "empno" in EMP table and "deptno" in DEPT table . I want to update both the columns with single UPDATE statement. With out a creation of stored procedure or view(updating it through view).
I M USING APEX 4.1 AND CREATED SELECT LIST ON PAGE, I WANT TO SHOW MIN VALUE OF THE SELECT LIST FOR THAT I WROTE IN THAT SELECT LIST PROPERTIES UNDER DEFAULT TAG MIN; AND CHOOSE PL/SQL EXPRESSION BUT ITS GIVING ERROR "Error computing item default value for page item P1_PRODUCT."
BUT IF I HARDCORE THE VALUE CONTAINING IN MY DATA LIKE PRODUCT ID = 1, I HARDCODED IN DEFAULT VALUE 1 AND SELECT PL/SQL EXPRESSION IT WORKS.
BUT ITS NOT DONE LIKE THIS I WANT TO SELECT BY DEFAULT MIN VALUE OF THE SELECT LIST, SO THAT THE DATA SHOULD BE DISPLAYED ACCORDING TO THAT.
THE EXACT REQUIREMENT IS TO ENTER THE SELECT LIST DEFAULT VALUE IN SESSION SO THAT DATA IS TO BE DISPLAYED.
I have an employee table that has a paygrp_id that will be used for my subset of employees. For all the employees that have the paygrp_id = 10212 on the employee table I need to update the workbrain_user table to set the flag wbu_cansee_self to 'N'. The join between the employee table and the workbrain_user table is the emp_id.I get the following error when I run this cursor.
[error] Error on line 0 DECLARE CURSOR wbuFlag_cur IS SELECT e.emp_fullname, e.paygrp_id,wbu.WBU_CANS [code]...
I have to update a table by getting values from two other tables. While doing that the inner query returns more than one value. I am not sure how to implement the logic without returning more than one row in sub query.
My query: update buf_office_str o set o.manager_ident = (select sp.ident from se2_r_src_sourceperson sp , (select distinct director_name, team_name from buf_sales_dump )t where SP.SRCNAME = upper(substr(t.director_name,instr(t.director_name,' ')+1,length(t.director_name))||', '||substr(t.director_name,1,instr(t.director_name,' ')-1 ) ) and o.office_descr = t.team_name )
Basically the query gets the manager id from sp table where sp.srcname = t.team-name. The office_descr should be equal to the team_name.
This is the logic I am working towards:
For each office, i get the office_descr and get corresponding team_name. Match the team's director_name (from table t) with the sp.name and return the employee's id (sp.ident) for that office_descr. I need to update all 50 offices with corresponding managerid for that office in buf_office_str table.
Is it possible to get done in one update? Or, does this need a plsql proc to do this logic.
ID DETAILS --- -------- 100 The ship has left san diego http:/localhost/icons/sandiego.png to okinawa on nov 10, 2011.
I need to update the record(id = 100) by replacing the url "http:/localhost/icons/sandiego.png" with "http:/localhost/icons/okinawa.png".
I need a procedure where I will pass the ID value, replace string(i.e http:/localhost/icons/sandiego.png) and replace with string (ie. http:/localhost/icons/okinawa.png).
In a pl/sql procedure, when I am doing an update, I need the old value to be returned and stored in a local variable, so that the same can be used for future purpose.
Note : I know the "OLD:" option is present when we use TRIGGER, but in my case , the table I am updating is a old table and I am not permitted to create a trigger for it.
