Multiple installs of causing Data provider internal error (-3000) on Open

Question

Hi,
We have an application written against Oracle ODP 10.1.0.401 that's been deployed out on a web farm working great for some time now, however recently another area has asked to get ODP for 11g installed on that server. The install originally broke our application, but we were able to resolve most of the issues after following all of the appropriate "PATH" reordering and registering appropriate dlls. In fact, we finally got a server completely back up and running both versions with no issues.
Unfortunately, it's a load-balanced server, and while one of the two servers is working correctly, the second seems to be having issues with specific parts of the install. We're trying to fix it on the second server now, and will be doing the same install shortly in our production environments.
I was able to break the error out and create a very simple test app to eliminate as many variables as possible. It appears to be an issue with the driver when we enlist into a Distributed Transaction.
The following code works fine:
var connection = new Oracle.DataAccess.Client.OracleConnection(connectionStringBase + "Enlist=False");
connection.Open();
While this code does not:
var connection = new Oracle.DataAccess.Client.OracleConnection(connectionStringBase);
connection.Open();
In both examples, connectionStringBase is Data Source=...;Validate Connection=true;User ID=...;Password=...; (where ... represents masked values). I've also tried "Enlist=True" in case the default was a problem, but it also fails.
The exceptions we're getting on the server happen in the .Open() call. The stack trace is:
[OracleException: Data provider internal error(-3000) [System.String]]
Oracle.DataAccess.Client.OracleException.HandleErrorHelper(Int32 errCode, OracleConnection conn, IntPtr opsErrCtx, OpoSqlValCtx* pOpoSqlValCtx, Object src, String procedure) +779
Oracle.DataAccess.Client.OracleException.HandleError(Int32 errCode, OracleConnection conn, IntPtr opsErrCtx, Object src) +41
Oracle.DataAccess.Client.OracleConnection.Open() +3338
Default.Page_Load(Object sender, EventArgs e) in C:\Code\Oracle11gTest\Oracle11gTest\Default.aspx.cs:13
System.Web.Util.CalliHelper.EventArgFunctionCaller(IntPtr fp, Object o, Object t, EventArgs e) +14
System.Web.Util.CalliEventHandlerDelegateProxy.Callback(Object sender, EventArgs e) +43
System.Web.UI.Control.OnLoad(EventArgs e) +91
System.Web.UI.Control.LoadRecursive() +74
System.Web.UI.Page.ProcessRequestMain(Boolean includeStagesBeforeAsyncPoint, Boolean includeStagesAfterAsyncPoint) +2604
I've read a few articles on this particular error and found that it's generally considered an ambiguous error with many potential causes, most of which I've been trying to test and rule out. While I don't have direct access to the server, I've been working with someone who does and they have compared registry settings and registered dlls, and have told me that the 2 servers match, and we've already re-installed the 10.1 drivers hoping it would help, but at this point we're stumped. I've read that it's possible that the COM+ install could have become corrupt, but what confuses me is that just prior to the 11g install, everything works great, and after doing no more than an ODP 11g install, it breaks. Do you have any advice or suggestions we could try?
Thanks in advance for any insight you might have!
~Jeff

Answer 1

If I read your post correctly that this is a "everything is equal as far as we know, but box A works and box B doesn't" type situation, my first suggestion would be to get Process Explorer from http://sysinternals.com, and compare the loaded modules between the working and non-working servers. You'll probably find that you're not using the software you think you're using, or have mismatched dlls across homes, or something to that effect. Once you figure out what the difference between the two is, correcting it will probably be the easy part.
Hope it helps,
Greg