Hi, I have a rather complex pyindi script that I have no problems to run manually, but when started on system boot (rc.local or systemd service), it usually fails (only occasionally/randomly it starts). Until lately, it worked (mostly) also on startup, but suddenly (after latest Astroberry updates?) it usually fails and I have to restart it manually.
It always fails with the following message:
Traceback (most recent call last):
File "/usr/local/lib/python3.7/dist-packages/PyIndi.py", line 1081, in <lambda>
__getattr__ = lambda self, name: _swig_getattr(self, BaseClient, name)
File "/usr/local/lib/python3.7/dist-packages/PyIndi.py", line 80, in _swig_getattr
raise AttributeError("'%s' object has no attribute '%s'" % (class_type.__name__, name))
AttributeError: 'BaseClient' object has no attribute 'removeDevice'
This exception is always preceded by a long list of messages of this type (just sometimes with different values or empty attributes):
Dispatch command error(-1): No device available and none was created
<defNumber name="%g" label="" format="%g" min="-32767" max="32767" step="0">
I cannot locate any particular place or operation triggering this exception. I have tried upgrading/downgrading pyindi-client between versions 0.2.5 and 0.2.7a (0.2.7 and 0.2.7b do not compile), but without success. Could it be some problem of the libindi 1.8.9 itself?
The cause seems to be a regular call to the infamous astroplan package, even just for what I thought a simple calculation. I have updated that package too and it does not seem to improve - rather the opposite.
I thought a delay in INDI event processing would cause missed events, but why does it cause this type of python exception?