Keep Coding

As a prophecy, I have a question related to a previous question of "cleaning" the database. How can I identify why my technosphere is not longer square?
I have done something to my database that if I try to do an LCIA of a random activity
def testactivity(activity):
method_key=methods.random()
fu={activity:1}
lca = LCA(fu,method_key)
lca.lci()
lca.lcia()
print(lca.score)
return()
testactivity(Database('ei_33consequential').random())
I get this warning message: NonsquareTechnosphere: Technosphere matrix is not square: 12384 activities (columns) and 12385 products (rows). Use LeastSquaresLCA to solve this system, or fix the input data.
I tried to find if I have a dataset with two reference products, to check that I looped through the database to check if the "production amount" was not a float. but I didn't find anything "wrong"
for ds in Database('ei_33consequential'):
if (isinstance(act['production amount'],float))==False:
print(ds['name'])
Is this approach correct to find an activity with more than one reference flow?. Otherwise, how can I find the product which is making my matrix non inversable?

You can check to see which activities have more than one production exchange with something like this:
for a in Database("ecoinvent 3.3 cutoff"):
assert len(a.production()) == 1