I have a procedure on Oracle that works perfectly fine if I call it from SQL Developer using this code:
VARIABLE x REFCURSOR
exec MY_PROCEDURE('par1', 'par2', 'par3', 'par4' ,:x);
PRINT x;
If I try to call it form my .Net app (using ODP.NET), I get the error:
Oracle.DataAccess.Client.OracleException ORA-08103: object no longer exists
This is the code I use to call it:
OracleConnection con = new OracleConnection();
con.ConnectionString = dbConnectionString; //string with the connectio. It is fine because I can connect
OracleCommand cmd = new OracleCommand("MY_PROCEDURE", con);
cmd.CommandType = CommandType.StoredProcedure;
cmd.Connection = con;
cmd.Parameters.Add(new OracleParameter("par1", OracleDbType.Varchar2)).Value = var1;
cmd.Parameters.Add(new OracleParameter("par2", OracleDbType.Varchar2)).Value = var2;
cmd.Parameters.Add(new OracleParameter("par3", OracleDbType.Varchar2)).Value = var3;
cmd.Parameters.Add(new OracleParameter("par4", OracleDbType.Varchar2)).Value = var4;
OracleParameter ref_cursor = new OracleParameter();
ref_cursor.OracleDbType = OracleDbType.RefCursor;
ref_cursor.Direction = ParameterDirection.Output;
cmd.Parameters.Add(ref_cursor);
con.Open();
OracleDataReader dr = cmd.ExecuteReader();
while (dr.Read())
{ ... }
The cmd.ExecuteReader command actually "works", the application exception is thrown on the dr.read but If I check the dr object, on the hasRows property I can see the ORA-08103: object no longer exists error.
What can be wrong?
One detail is that I have a similar procedure that follows pretty much the same logic (of returing a cursor) and works fine.
Does the query for the returned cursor involve temporary tables? You can shoot yourself in the foot if you return a cursor involving temporary tables with the ON COMMIT DELETE ROWS options and then commit the transaction before you have retrieved the cursor data.
The COMMIT easily happens because ODP.NET by default works in auto commit mode.
To fix it,
either turn off auto commit,
or use temporary tables with the ON COMMIT PRESERVE ROWS options (instead of ON COMMIT DELETE ROWS),
or use regular tables.
You can also create a Transaction for your connection, and set the transaction in the OracleCommand object.
It will also keep the commits from happening before you retrieve all the data.
Related
Scenario
I'm working with SQL Server 2017 (not possible to change)
I'm using Visual Studio 2019 in C# console and .NET Framework 4.5 (possible to change)
I'm using ADO.NET because several years before we couldn't use Entity Framework, as the system is made to work with a stored procedure that returns at least 100k rows (possible to change)
Situation
I have an USP that returns a table that is at least 100k of rows by 20 fields. I need to add an output parameter in order to get also an ID created by the USP itself. So, the situation is that I need return a table and an ID (called ProcMonitorId). I don't know if this is even so possible (See workarounds section)
At the SQL level is seems to be so far so good:
CREATE PROCEDURE [myschema].[mystore]
#ProcMonitorId BIGINT OUTPUT
AS
BEGIN
BEGIN TRANSACTION
(...)
SELECT fields FROM myTable
SELECT #ProcMonitorId = #internalVariable
SQL execution:
And at repository layer (only relevant lines, someone were surprised for health of example):
var command = new SqlCommand("myStoreProcedure", mycon);
command.CommandType = CommandType.StoredProcedure;
SqlParameter outPutParameter = new SqlParameter();
outPutParameter.ParameterName = "#ProcMonitorId";
outPutParameter.SqlDbType = System.Data.SqlDbType.BigInt;
outPutParameter.Direction = System.Data.ParameterDirection.Output;
command.Parameters.Add(outPutParameter);
// Open connection etc-etc that works
SqlDataAdapter da = new SqlDataAdapter(command);
DataTable dt = new DataTable();
string ProcMonitorId = outPutParameter.Value.ToString();
da.Fill(dt);
Everything worked fine until the addition of the output at C# level. It returns in the line:
string ProcMonitorId = outPutParameter.Value.ToString();
it returns NullReferenceException because Value is null (that can't be) and of course, can't convert to String. I would solve this situation by adding a ? but if that's situation happens for real, I need catch it any way as error. The main idea is that Value can not be null.
As I don't have any ORM map, (and my expertise is not ADO.NET but Entity Framework) I can't understand why is null (No, is not null at SQL layer, always return a value)
Question
How can I solve this error or how can I return a BIGINT parameter and ALSO a table result?
Workarounds
As I first glance I have to solve it quickly, I made a:
SELECT 1 as type, #procID as procid, null as data1, null as data2
UNION ALL
SELECT 2 as type, null as procid, data1, data2
in order to simulate a "header" and "data" rows on one single table.
But I don't like this solution and is not very elegant and flexible. I've to parse the header every time.
Thanks in advance and please comment anything, tip, help, workaround, I will be glade to update my answer if more information is needed.
Also I can make my Framework to .NET Core or change to Entity Framework. That I can't change is my SQL version
Update #2
No changes in SQL - Still working as screenshot
In C# - Hangs out for ever
SqlConnection connection = new SqlConnection(ConfigurationManager.AppSettings["DbConnection"]);
connection.Open();
var command = new SqlCommand("myUSP", connection);
command.CommandType = CommandType.StoredProcedure;
command.CommandTimeout = Convert.ToInt16(ConfigurationManager.AppSettings["DataBaseTimeOut"]);
if (connection.State != ConnectionState.Open)
{
connection.Open();
}
SqlParameter r = command.Parameters.Add("#ProcMonitorId", SqlDbType.BigInt);
r.Direction = ParameterDirection.Output;
DataTable dt = new DataTable();
using (var rdr = command.ExecuteReader())
{
dt.Load(rdr);
long id = (long)r.Value;
}
SqlDataAdapter da = new SqlDataAdapter(command);
da.Fill(dt);
The parameter value won't be available until after you consume the resultset, eg
var cmd0 = new SqlCommand("create or alter procedure pFoo #id int output as begin select * from sys.objects; set #id = 12; end", con);
cmd0.ExecuteNonQuery();
var cmd = new SqlCommand("pFoo", con);
cmd.CommandType = CommandType.StoredProcedure;
var p1 = cmd.Parameters.Add("#id", SqlDbType.Int);
p1.Direction = ParameterDirection.Output;
var dt = new DataTable();
using (var rdr = cmd.ExecuteReader())
{
dt.Load(rdr);
var id = (int)p1.Value;
}
You should use a Parameter with the Direction property set to ReturnValue, and, inside the sp, declare an internal variable and set it to the value you want.
Then call the RETURN statement before leaving the StoredProcedure.
As an example, see this SP:
ALTER PROCEDURE [GetTimeZoneGMT]
#TimeZone NVARCHAR(128)
AS
BEGIN
DECLARE #timeZoneNumber as INT = -20;
IF #TimeZone ='Pacific/Midway'
SET #timeZoneNumber = -11
ELSE IF #TimeZone ='Pacific/Niue'
SET #timeZoneNumber = -11
ELSE IF #TimeZone ='Pacific/Pago_Pago'
SET #timeZoneNumber = -11
SELECT 1 -- or whatever you need to have as result set
RETURN #timeZoneNumber;
END
The stored procedure ends with a (bogus) SELECT statement but also has a RETURN statement with the parameter set inside the SP logic.
Now from the C# side you could call it in this way (LinqPad example)
using (var connection = new SqlConnection("Data Source=(LOCAL);Initial Catalog=LinqPADTest;Integrated Security=True;"))
{
connection.Open();
SqlCommand cmd = new SqlCommand("GetTimeZoneGMT", connection);
cmd.CommandType = CommandType.StoredProcedure;
cmd.Parameters.Add("#TimeZone", SqlDbType.NVarChar).Value = "Asia/Kuala_Lumpur";
SqlParameter r = cmd.Parameters.Add("#p2", SqlDbType.BigInt);
r.Direction = ParameterDirection.ReturnValue;
DataTable dt = new DataTable();
dt.Load(cmd.ExecuteReader());
r.Value.Dump(); // Prints -20
dt.Dump(); // Prints a row with a single column with 1 as value
}
From C# Code, I'm trying to call a PACKAGE.PROCEDURE() from Oracle. In this simple example I should get one value from the procedure call, but all I get is error:
wrong number or types of arguments in call to 'RETURN_NUM'
The procedure is declared as follows:
PROCEDURE return_num(xNum OUT NUMBER) AS
BEGIN
xNum:= 50;
dbms_output.put_line('hello world ' || xNum);
END;
C# code:
Oraclecon.Open();
OleDbCommand myCMD = new OleDbCommand("TEST.return_num", Oraclecon);
myCMD.CommandType = CommandType.StoredProcedure;
myCMD.Parameters.Add("xNum", OleDbType.Numeric);
OleDbDataReader myReader;
myReader = myCMD.ExecuteReader();
Can some one please point out what I'm doing wrong. Then in a real scenario I would like to call a procedure that returns a set of values from a custom Type, such as:
TYPE r_interface_data IS RECORD
(
object_id VARCHAR2(16),
obj_type VARCHAR2(32)
);
TYPE t_interfase_data IS TABLE OF r_interface_data;
How can I approach that. Thanks!
UPDATE: In my particular case I ended-up doing the following approach
using (OleDbCommand cmd = new OleDbCommand("PACKAGE.procedure_name"))
{
cmd.CommandType = CommandType.StoredProcedure;
SqlManager sqlManager = new SqlManager();
return sqlManager.GetDataSet(cmd);
}
I don't think you're that far off... try this:
OracleCommand cmd = new OracleCommand("return_num", Oraclecon);
cmd.Parameters.Add(new OracleParameter("xNum", OracleDbType.Decimal,
ParameterDirection.Output));
cmd.CommandType = CommandType.StoredProcedure;
cmd.ExecuteNonQuery();
OracleDecimal d = (OracleDecimal)cmd.Parameters[0].Value;
double result = d.ToDouble();
result now contains the out parameter from the procedure.
I think your problem is you were attempting to use a DbDataReader on a stored procedure. DbDataReader is for queries.
Also, I used ODP.net -- that may or may not have contributed to your issue, that you were using Ole.
I'm working on one program that I need to modify a little. There's one SQL statement I don't understand what it does (or basically how it does it).
string query = "SELECT dbo.BusinessMinutes(#start,#end,#priorityid)";
SqlCommand cmd = new SqlCommand(query, con);
cmd.Parameters.Add("#start", SqlDbType.DateTime).Value = start;
cmd.Parameters.Add("#end", SqlDbType.DateTime).Value = end;
cmd.Parameters.Add("#priorityid", SqlDbType.UniqueIdentifier).Value = priorityId;
SqlDataAdapter READER = new SqlDataAdapter();
READER.SelectCommand = cmd;
DataTable table = new DataTable();
READER.Fill(table);
if (table.Rows.Count == 1)
{
minutes = (int)table.Rows[0][0];
}
So can someone explain me the SELECT statement there. The end result (minutes) is as expected so it works but that syntax confuses me. Is this somehow equal to SELECT * FROM dbo.BusinessMinutes WHERE...
Is this commonly used and does this syntax has some special name so I could name my question better? Thank you in advance.
dbo.BusinessMinutes has to be a UDF (User Defined Function) that returns a simple scalar value based on a starting date, an ending date and a priority indicator.
Scalar functions, be it UDF or native, can be used in a SELECT statement to produce a field in the returned resultset. Thus, the code you have is perfectly legal.
For more information about scalar UDF, read this MSDN article.
As a side note, a better implementation for that code would be this:
string query = "SELECT dbo.BusinessMinutes(#start,#end,#priorityid)";
using (SqlCommand cmd = new SqlCommand(query, con))
{
cmd.Parameters.Add("#start", SqlDbType.DateTime).Value = start;
cmd.Parameters.Add("#end", SqlDbType.DateTime).Value = end;
cmd.Parameters.Add("#priorityid", SqlDbType.UniqueIdentifier).Value = priorityId;
// assuming connection is already open
using (SqlDataReader reader = cmd.ExecuteReader())
{
if (reader.Read()) minutes = reader.GetInt32(0);
}
}
If all you want is to init the minutes variable, using SqlDataReader will be more efficient and performant than creating a SqlDataAdapter and a DataTable. The using statements will also make sure your SqlCommand and SqlDataReader objects gets disposed of properly.
It is not a table name. I think you call into a FUNCTION.
how to create and call scalar function in sql server 2008
has more explanations and examples.
This is a simple task that I want to acheive but ASP.NET makes it quite difficult, next to impossible. I followed this question
Running a Stored Procedure in C# Button but found out ExecuteNonQuery does not return the output from query.
I tried this other approach but can't seem to pass the paremeters in this way
SqlConnection myConnection = new SqlConnection(myconnectionString);
SqlCommand myCommand = new SqlCommand();
myCommand.CommandType = CommandType.StoredProcedure;
myCommand.CommandText = "usp_GetCustomer";
myCommand.SelectParameter <-- does not exist
Can someone write this simple code, how can I implement it? Basically I am passing a #username and #month (both character strings) to stored procedure and it returns a number that I want to capture and assign to a label control.
Thank you
The output from my query is this. It runs a complex query, create a temp table and then it runs
select ##rowcount
and I am capturing that.
Don't use SqlCommand.ExecuteNonQuery() if you actually want data from a result set.
Make sure your procedure uses set nocount on
Then use SqlCommand.ExecuteScalar()
return (int)myCommand.ExecuteScalar(); // value of select ##rowcount
Edit: As for your parameters:
myCommand.Parameters.AddWithValue("#username","jsmith");
myCommand.Parameters.AddWithValue("#month","January");
I prefer using linq-to-sql to handle stored procedures. Create a linq-to-sql model, where you add the SP you want to call. This will expose the SP as a function on the generated data context, where the parameters are ordinary C# functions. The returned values will be exposed as a collection of C# objects.
If you have multiple results from the SP things get a bit more complicated, but still quite straight forward.
Use the Parameters collection of the command to set the parameters, and the ExecuteScalar to run the query and get the value from the single-row single-column result.
Use using blocks to make sure that the connection and command are closed and disposed properly in any situation. Note that you have to provide the connection to the command object:
int result;
using (SqlConnection connection = new SqlConnection(myconnectionString)) {
using (SqlCommand command = new SqlCommand(connection)) {
command.CommandType = CommandType.StoredProcedure;
command.CommandText = "usp_GetCustomer";
command.Parameters.Add("#username", SqlDbType.VarChar).Value = username;
command.Parameters.Add("#month", SqlDbType.VarChar).Value = month;
connection.Open();
result = (int)myCommand.ExecuteScalar();
}
}
using(SqlConnection myConnection = new SqlConnection(myconnectionString))
{
SqlCommand myCommand = new SqlCommand();
myCommand.CommandType = CommandType.StoredProcedure;
myCommand.CommandText = "usp_GetCustomer";
myCommand.Parameters.Add("#USER_NAME", SqlDbType.VarChar).Value = sUserName; // user name that you pass to stored procedure
myCommand.Parameters.Add("#Month", SqlDbType.VarChar).Value = iMonth; // Month that you pass to stored procedure
// to get return value from stored procedure
myCommand.Parameters.Add("#ReturnValue", SqlDbType.Int).Direction = ParameterDirection.ReturnValue;
myConnection .Open();
myCommand.ExecuteScalar();
// Returnvalue from stored procedure
return Convert.ToInt32(command.Parameters["#ReturnValue"].Value);
}
Simple code to get return value from SQL Server
SqlConnection myConnection = new SqlConnection(myconnectionString);
SqlCommand myCommand = new SqlCommand();
myCommand.CommandType = CommandType.StoredProcedure;
myCommand.CommandText = "usp_GetCustomer";
myCommand.Parameters.Add("#USER_NAME", SqlDbType.VarChar).Value = sUserName; // user name that you pass to the stored procedure
myCommand.Parameters.Add("#Month", SqlDbType.VarChar).Value = iMonth; //Month that you pass to the stored procedure
// to get return value from the stored procedure
myCommand.Parameters.Add("#ReturnValue", SqlDbType.Int).Direction = ParameterDirection.ReturnValue;
myConnection .Open();
myCommand.ExecuteScalar();
// Returnvalue from the stored procedure
int iReturnValue = Convert.ToInt32(command.Parameters["#ReturnValue"].Value);
I am executing a stored procedure that has no return value. How can I check that it has actually been executed? Here is the code:
this.dbProviderFactory = DalFactory.GetFactory(this.adapterConfiguration);
DbConnection dbConnection = dbProviderFactory.CreateConnection();
dbConnection.ConnectionString = this.adapterConfiguration.DatabaseInformation.ExternalDatabaseInformation.connectionString;
dbConnection.Open();
DbCommand cmd = dbConnection.CreateCommand();
cmd.CommandText = "h_AS_SP_ResetUnfinishedJobs";
cmd.CommandType = CommandType.StoredProcedure;
cmd.ExecuteNonQuery();
And here is the stored procedure:
ALTER PROCEDURE [dbo].[h_AS_SP_ResetUnfinishedJobs]
AS
BEGIN
-- Delete all unfinished jobs where the force flag has not been set...
DELETE FROM h_AS_mds_MetaDataStatus
WHERE mds_status NOT IN (11,12) AND mds_force = 0
END
The stored proc will return a "number of rows affected" when using ExecuteNonQuery():
DbCommand cmd = dbConnection.CreateCommand();
cmd.CommandText = "h_AS_SP_ResetUnfinishedJobs";
cmd.CommandType = CommandType.StoredProcedure;
int rowsAffected = cmd.ExecuteNonQuery();
This will give you an idea whether or not anything has even been done. However: if not affecting any rows also is a valid outcome for your stored proc, you cannot really use that return value to check if it's been run.
Other than that: unless an exception occurs, the call presumably worked !