Re: WebLogic RDB jdbc problem

Question

Hi Joe,
I tried the two things you asked:
1. Setting the initial and max capacity of the Connection Pool to 1 does not make
any further difference, the reselt set is still sometimes empty.
2. If I use the code from the stand-alone application (thus if I get the connection
directly from the driver) than I get a good result: the retrieved result set is,
as far as I can see, always correctly filled.
In addition I tried to cast the Connection to WLConnection and then get the vendor
connection from it, and than to proceed with this connection further on, but this
did not make it any better.
Almir.
Joe Weinstein <[email protected]> wrote:
>
>
Almir wrote:
Hi Joe,
thanks for your quick answer to my questions. I red your sugestionsand tried
them out but unfortunately it did not make me come any further in searchfor answer
to this odd problem. Setting the statement cache size to 0 (I've alreadytried
this before your response) did not make any difference at all. Refreshingthe
statement before reuse neighter.
I also tried to make the stand-alone application to go wrong. But nomatter whether
I reuse connection and/or statement or whether I instantiate the newones every
time, the execution of the statement always comes back with a validand filled
ResultSet.
Interesting thing to mention aboout executing the statement withinthe server
is that if I put it in a loop of let say 100 repeats, it succeeds orit fails
all 100 times. Next time i execute this bean method where the loopis situated
it will again succeed or fail all 100 times. It remains consequentwithin the
same loop/bean method. Even if I get the data source and the connectionevery
time again (100 times) it remains consequent. Thi is maybe interestingfinding
but it does not say anything to me. Perhaps you can make some conclusionsout
of it.
Best regards,
Almir.Hmmmmm.... Interesting. The only other thing I can suspect is that the
driver
is not threadsafe... Would you try these two experiments?
1 - Alter the weblogic server to define the pool to have only one connection,
ie: initial capacity = 1 and max capacity = 1. Let me know if the
problem
is still there.
2 - Alter the application code you run in the server to get a connection
directly
from the driver, just like the standalone code.
Let me know...
Joe
Joe Weinstein <[email protected]> wrote:
Hi Almir.
I highly suspect the RDB driver of a bug where re-using a given Prepared
Statement
will sometimes produce an empty result set. Weblogic will cache your
prepared statements
so when you call prepareStatement() we may transparently return you
a
previously-used
statement for the same SQL and arguments.
To prove my theory, please do these two things:
1 - Define your pool's statement cache to be of size zero. This should
make the problem
go away.
2 - In a standalone program, you can create a given PreparedStatement,
and rerun it in
a loop, and see if it ever returns an empty result set. I will showyou
how to include
the code we do to refresh the statement before reuse:
try
Class.forName ("oracle.rdb.jdbc.rdbThin.Driver");
Connection conn = DriverManager.getConnection("jdbc:rdbThin://<database_name>","<user>",
"<pass>");
String sqlStatement = "SELECT TABEL1.COLUMN1, TABEL1.COLUMN2,
TABEL2.COLUMN1
FROM TABEL1, TABEL2 WHERE (TABEL2.COLUMN2= TABEL1.COLUMN3) AND (TABEL2.COLUMN1
= ?)ORDER BY TABEL1.COLUMN1";
PreparedStatement ps = conn.prepareStatement(sqlStatement,
ResultSet.TYPE_SCROLL_INSENSITIVE,ResultSet.CONCUR_READ_ONLY);
for (int i = 0; i < 100; i++)
ps.setObject(1, new Integer(1));
boolean result = ps.execute();
ResultSet rs = ps.getResultSet();
// check result set here
rs.close();
cleanUpStatementForReUse(ps);
ps.close();
conn.close();
catch (Exception ex)
System.out.println(ex);
private void cleanUpStatementForReUse(PreparedStatement p)
try { p.clearParameters(); p.clearBatch(); } catch (Throwable ignore)
try { p.setEscapeProcessing(true); } catch (Throwable ignore) {}
try
if (p.getFetchDirection() != ResultSet.FETCH_FORWARD)
p.setFetchDirection(ResultSet.FETCH_FORWARD);
} catch (Throwable ignore) {}
try
if (p.getFetchSize() != 0)
p.setFetchSize(0);
} catch (Throwable ignore) {}
try
if (p.getMaxFieldSize() != 0)
p.setMaxFieldSize(0);
catch (Throwable ignore) {}
try
if (p.getMaxRows() != 0)
p.setMaxRows(0);
catch (Throwable ignore) {}
try
if (p.getQueryTimeout() != 0)
p.setQueryTimeout(0);
catch (Throwable ignore) {}
try
p.clearWarnings();
catch (Throwable ignore) {}
Almir wrote:
Hello everybody,
we are developing an application for Bea WebLogic Application Server8.1. This
application uses data from an existing database. This is an OracleRDB 7-1.220
database. Because the WebLogic AS does not include a JDBC driver forthis type
of database we got one from Oracle: RDBJDBC71. I installed this driverand tested
it by using a standalone Java application which executes a number
of
SQL statemens.
This test was satisfying, no problems occured.
Then, in order to use it within the application server, I defined
a
JDBC Connection
Pool in WebLogic with corresponding Data Source. And it worked, ourapplication
could connect with the database and execute the needed transactions.
But after a while we noticed a realy strange problem. When a particularSELECT
SQL stament is executed we expect a list of 6 records as result. Butsometimes
the returned resultset is just empty. I tried this hundreds of times.No error
or exception occurs but sometimes the response is correct and sometimesnot.
In the first place we thought that the JDBC driver was causing thisproblem but
then I executed the same SQL Select statement within a Java stand-aloneapplication
(and not using the Application Server) and the result was correct.Repeating this
hundreds of times did not change the outcome, the given resultset
was
always correct.
I tried reconfigurating the Connection Pool in different ways but
it
did not make
any difference: sometimes it worked and sometimes it didn't.
Here below you can see the code samples of both implementations.
Stand-alone:
try
Class.forName ("oracle.rdb.jdbc.rdbThin.Driver");
Connection conn = DriverManager.getConnection("jdbc:rdbThin://<database_name>","<user>",
"<pass>");
String sqlStatement = "SELECT TABEL1.COLUMN1, TABEL1.COLUMN2,TABEL2.COLUMN1
FROM TABEL1, TABEL2 WHERE (TABEL2.COLUMN2= TABEL1.COLUMN3) AND (TABEL2.COLUMN1
= ?)ORDER BY TABEL1.COLUMN1";
PreparedStatement ps = conn.prepareStatement(sqlStatement,ResultSet.TYPE_SCROLL_INSENSITIVE,ResultSet.CONCUR_READ_ONLY);
ps.setObject(1, new Integer(1));
boolean result = ps.execute();
ResultSet rs = ps.getResultSet();
// Result set is here always filled!!!!
rs.close();
ps.close();
conn.close();
catch (Exception ex)
System.out.println(ex);
Application server:
try
Context ctx = new InitialContext();
Object object = ctx.lookup(data_store_name);
DataStore ds = (DataStore)object;
Connection conn = ds.getConnection();
String sqlStatement = "SELECT TABEL1.COLUMN1, TABEL1.COLUMN2,TABEL2.COLUMN1
FROM TABEL1, TABEL2 WHERE (TABEL2.COLUMN2= TABEL1.COLUMN3) AND (TABEL2.COLUMN1
= ?)ORDER BY TABEL1.COLUMN1";
PreparedStatement ps = (PreparedStatement)conn.prepareStatement(sqlStatement,
ResultSet.TYPE_SCROLL_INSENSITIVE,ResultSet.CONCUR_READ_ONLY);
ps.setObject(1, new Integer(1));
boolean result = ps.execute();
ResultSet rs = ps.getResultSet();
// Result set is here sometimes empty !!!!
rs.close();
ps.close();
conn.close();
catch (Exception ex)
System.out.println(ex);
Any sugestions are welcome.
Thanks.
Almir.

Answer 1

I have had the same problem as you with sporadic results for queries with RDB Drivers. Were you able to solve it and if so how?
Sam Opoku-Boadu