In TensorFlow v2, below code can cause GraphDef reconciliation error.

@tf.function
def foo(x):
  return x ** 2

with writer.as_default():
  tf.summary.trace_on()
  foo(1)
  foo(2)
  tf.summary.trace_export("foo")

Depending on the argument, tf.function (really, auto-graph) creates ops that are unique within GraphDef but is not globally unique. In the example above, two GraphDefs (on from foo(1) and another from foo(2)) will be written out and they can collide badly in names and content.

In such case, instead of showing wrong graph content, TensorBoard throws an error.