I have a loop which seems to work for some time but then after a number of messages it starts to fall through even there is no data available on the Reader.

Basically,

j = journal.Reader()
j.this_boot()
poller = select.poll()
poller.register(j.fileno())
while True:
    poller.poll()  # returns within 1E-6 seconds
    print(j.get_next())  # prints out {}

Is this some fundamental use error on my side? The poller object should only return when the journal endpoint has some data to be read?

I have systemd-python version:

In [6]: systemd.journal.__version__
Out[6]: '234'