Hello, all.

We have deployed an ODP.NET-based procedure (10.2.0.2) within our SqlServer 2005 environment for some custom data retreival back from some particular Oracle servers. In general, this works well. However, after a period of time, this remote fetch system fails, reporting that it is unable to allocate memory while executing the Read method of the OracleDataReader in use. The only solution seen so far is to restart SqlServer.

After considerable research on our database server in conjunction with our local DBA, our analysis of our database server shows it is not under heavy memory pressure when this error occurs. Beyond that, we've come to the conclusion that the memory error is a bit of a red herring. In reality, we believe some downstream process within the ODP.NET structure is, for lack of a more technically accurate term, misbehaving, and the failure is being misinterpreted as a memory allocation failure. One dump of data from a failed process in relation to this offered the famous "Not enough storage to proces this command," which is *very* often linked to Windows RPC failures.

On that basis, I have become *very* interested in the relationship between the ODP relationship to the Windows RPC server. I have a rough speculation that there is some presently unknown set of circumstances that, after a time, cause the Oracle providers to fail in establishing whatever remote connections are necessary to facilitate the flow of data from the Oracle server back to our SQL client. I strongly suspect the RPC is a player in this, hence the question.

Anyone that might be able to provide some insight (or even some known issues) regarding ODP's interaction with RPC would be greatly appreciated.


Many thanks,
SoonerDave

0  out of 1 found this valuable. Do you find this valuable? Do you?    

• Post reply