I am trying to use django-dramatiq
to run a pipeline of several stages, each defined as dramatiq Actor
s, using the pipeline(<stages>).run()
method but it's only running the first stage/Actor
and not attempting the other stages.
I've defined some cutdown fake actors to illustrate the problem:
import dramatiq
@dramatiq.actor
def fake_extract(process_pk, *args, **kwargs):
print(f"fake_extract: Process PK= {process_pk} Running extract on {kwargs['fits_file']}")
@dramatiq.actor
def fake_astromfit(process_pk, *args, **kwargs):
print(f"fake_astromfit: Process PK= {process_pk} Astrometric fit on {kwargs['ldac_catalog']}, updating {kwargs['fits_file']}")
@dramatiq.actor
def fake_zeropoint(process_pk, *args, **kwargs):
print(f"fake_zeropoint: Process PK= {process_pk} ZP determination on {kwargs['ldac_catalog']} with {kwargs['desired_catalog']} ref catalog")
I've then defined the stages and built a pipeline:
import os
from dramatiq import pipeline
from test_dramatiq.dramatiq_tests import fake_extract, fake_astromfit, fake_zeropoint
fits_filepath = '/foo/bar.fits'
fits_file = os.path.basename(fits_filepath)
steps = [{
'name' : 'proc-extract',
'runner' : fake_extract,
'inputs' : {'fits_file':fits_filepath,
'datadir': os.path.join(dataroot, temp_dir)}
},
{
'name' : 'proc-astromfit',
'runner' : fake_astromfit,
'inputs' : {'fits_file' : fits_filepath,
'ldac_catalog' : os.path.join(dataroot, temp_dir, fits_file.replace('e91.fits', 'e91_ldac.fits')),
'datadir' : os.path.join(dataroot, temp_dir)
}
},
{
'name' : 'proc-zeropoint',
'runner' : fake_zeropoint,
'inputs' : {'ldac_catalog' : os.path.join(dataroot, temp_dir, fits_file.replace('e91.fits', 'e92_ldac.fits')),
'datadir' : os.path.join(dataroot, temp_dir),
'desired_catalog' : 'PS1'
}
}]
pipes = []
for step_num, step in enumerate(steps):
inputs = step['inputs']
print(f" Performing pipeline step {step['name']}")
pk = 1234+step_num
pipes.append(step['runner'].message_with_options(args=[pk,], kwargs=inputs, pipe_ignore=True))
pipeline(pipes).run()
Running inside ipython
with regular dramatiq
these seems to work fine and all stages run:
fake_extract: Process PK= 1234 Running extract on /foo/bar.fits
fake_astromfit: Process PK= 1235 Astrometric fit on /foo/Temp_cvc2/bar.fits, updating /foo/bar.fits
fake_zeropoint: Process PK= 1236 ZP determination on /foo/Temp_cvc2/bar.fits with PS1 ref catalog
However defining them in a module imported by django-dramatiq
via the Django project's settings.py
file and defining the pipeline as above in the python manage.py shell
and with the python manage.py rundramatiq
runner only runs the first stage/Actor
:
fake_extract: Process PK= 1234 Running extract on /foo/bar.fits
and it never executes the other stages...
Any ideas on what is going on here and why multi-stage pipelines are not working under django-dramatiq
?