pg_trace() enables tracing of
the PostgreSQL frontend/backend communication to a debugging
file specified as pathname.
To fully understand the results, one needs to be familiar
with the internals of PostgreSQL communication protocol. For
those who are not, it can still be useful for tracing errors
in queries sent to the server, you could do for example grep '^To backend' trace.log and see what
query actually were sent to the PostgreSQL server. For more
information, refer to PostgreSQL manual.
pathname and mode are the same as in
fopen() (mode
defaults to 'w'),
connection specifies the connection to trace and
defaults to the last one opened.
It returns TRUE if pathname could be opened for
logging, FALSE
otherwise.
See also fopen() and
pg_untrace().