DTN IQFeed, DTN.IQ, DTN ProphetX Support Forums

Hello, what you seem to be experiencing happens by design (and is not something new).

When the server gets an error on the client socket, it will drop the data rather than queue it for a later send. This is designed this way so that no single client can adversly affect other clients connected to the server.

On the other hand, IQConnect is designed so that it will queue data that gets received from the server indefinately until it gets delivered to the client app or until the connection is dropped.

What this means is that for the most part, the data processing in your app and the operating envionment for IQConnect are responsible for problems such as these. If your app is slow reading data off the socket, then IQConnect is going to be queing data internally. The more time IQConnect has to spend managing it's internal data queues, the less time it has available to read from the server socket and eventually the socket buffer on the serverside will fill up causing errors on the socket and the server to drop data. This is why it takes a few minutes for the problem to occur.

You mentioned that you didn't experience these problems with 4.4.0.3. My initial reaction would be that you simply didn't notice the problem with that version. Also keep in mind that these types of problems are also greatly dependant upon the market conditions so it is possible that your current issue wouldn't have existed during your testing of 4.4.0.3 becuase there was less data comming through the system. Our internal testing indicated that 4.5.0.3 is significantly more efficient at processing the streaming quotes feed than 4.4.0.3. However, this has the side effect of your app getting delivered larger "bursts" of messages and the potential to end up queuing data internally is greater since more data is actually getting delivered to your app in a shorter amount of time.

Of course with your setup, you also have the added (untested) overhead of running IQFeed in wine. I am not sure how much overhead this adds to the system but it should be significant in relation to IQFeed.

Hopefully this explanation gives you a better idea of what is going on. As I mentioned above, the only real control you have as a developer is increasing the efficiency of your data processing within your app which will reduce the load on the IQConnect client running on your computer.

Other than that, it is possible (although not probable) that you are doing something with the feed we hadn't anticipated and simply changing some methodology might help as well. Is there anything unusual about how you are watching these symbols (do you use fieldsets or do you make multiple connections to IQconnect to do so)?