I'm not sure if this behavior is the desired one or not, but simply specifying nn, nn_port, rm, or rm_port does not pass this along to Knit. I'm guessing this is not desired b/c of the comment. Starting at line 47 in core.py:
# if any hdfs/yarn settings are used don't use autodetect
if autodetect or not any([nn, nn_port, rm, rm_port]):
self.knit = Knit(autodetect=True, validate=validate)
which ignores the settings unless the default True value for autodetect is overridden.
I found it very confusing that simply passing the nn and rm args were not enough to have those pass through to Knit, but I also had to add autodetect=False.
I'm not sure if this behavior is the desired one or not, but simply specifying
nn
,nn_port
,rm
, orrm_port
does not pass this along to Knit. I'm guessing this is not desired b/c of the comment. Starting at line 47 incore.py
:which ignores the settings unless the default
True
value forautodetect
is overridden. I found it very confusing that simply passing the nn and rm args were not enough to have those pass through to Knit, but I also had to addautodetect=False
.Is this the desired behavior or a bug?