Mamta Satoor <[email protected]> writes:
> Hi,
>
> I am working on writing a test for INSERT statement which will cause
> INSERT trigger to fire. The INSERT is being done into a table with a
> large BLOB column. The table definition looks as follows create table
> table1 (id int, status smallint, bl blob(2G))
>
> My goal is to run this INSERT statement inside a java program with a
> limited heap to see if I will run into any OOMs during trigger
> execution because we are trying to stream the data into memory. The
> way I am trying to insert large LOB is as follows
> System.out.println("Inserting into table1 to cause insert
> trigger to fire");
> PreparedStatement ps = conn.prepareStatement(
> "insert into table1(id, status, bl) values(101, 0, ?)");
> byte[] arr = new byte[300000*1024];
> for (int i = 0; i < arr.length; i++)
> arr[i] = (byte)4;
> ps.setBinaryStream(1, new ByteArrayInputStream(arr),
> arr.length);
> ps.executeUpdate();
> conn.commit();
> But the above code runs into OOM for byte[] arr = new
> byte[300000*1024]; because I do not have enough heap available. I was
> wondering if there was a way for me to use a stream to insert a pretty
> large LOB into my table without having a physical file on the disk
> from which I will stream the data in. I do not care about the actual
> data. As shown above, I am just inserting a large quantity of (byte)4
> in the LOB. Would appreciate if anyone knows of a way for me to stream
> large data into LOB while running with limited heap.
You might use the same method as used in BlobMemTest, perhaps?
(LoopingAlphabetStream).
Thanks,
Dag
