PRB: Restarting a Paused Profiler Trace May Return General Network Error (272706)

SYMPTOMS

If the server against which a trace is being run is restarted, while the trace is paused, the profiler returns the following series of error messages in a dialog box:

Failed to restart a trace.
[Microsoft][ODBC SQL Server Driver][Shared Memory]ConnectionWrite
(WrapperWrite()). (State 01000) (Code 55)
[Microsoft][ODBC SQL Server Driver][Shared Memory]General network error. Check your network documentation. (State 08S01) (Code 11)

The information window also displays the following:

Error retrieving trace information from the server

as well as:

Connections: 0

However, it may not not be apparent to the user that the network error messages are occurring because of a server restart.

Following the reporting of the error messages, the Pause and Stop Execution buttons are dimmed. Only the Start execution button is enabled. Restarting the trace works successfully, with the exception that all previously obtained trace results are cleared from the window and a new trace (and the resulting output) are begun from scratch.

CAUSE

This behavior is considered to be by design. Several different ideas were considered in order to make this behavior more intuitive, including having the profiler detect for a dead connection before attempting to restart the trace. Ultimately, it was decided to have the ODBC driver propagate the exact error messages to the user interface (UI). The current behavior enables the user to be made aware of the fact that something has gone wrong with the connection. However, the user needs to take care to save any existing results before restarting the trace. Both saving to a table and saving to a file work properly.