Closed martinschorb closed 2 years ago
You are using an outdated version of z5py
. Please update it then this will work.
Also, like in mobie/mobie-utils-python#57 I noticed the missing
str()
here:
Yes, that's a good point, can you make a PR?
You are using an outdated version of
z5py
OK, I re-installed cluster_tools
and mobie
through pip from the git repos but that only checks for existence and does not update dependencies.
Yes, that's a good point, can you make a PR? https://github.com/constantinpape/cluster_tools/pull/35
Now with everythig updated I get:
...
...cluster_tools/copy_volume/copy_volume.py", line 162, in run_impl
self.submit_jobs(n_jobs, self.prefix)
File ".../cluster_tools/cluster_tasks.py", line 567, in submit_jobs
tasks = [pp.submit(self._submit, job_id, job_prefix) for job_id in range(n_jobs)]
File ".../cluster_tools/cluster_tasks.py", line 567, in <listcomp>
tasks = [pp.submit(self._submit, job_id, job_prefix) for job_id in range(n_jobs)]
File ".../python3.8/concurrent/futures/process.py", line 645, in submit
self._start_queue_management_thread()
File ".../python3.8/concurrent/futures/process.py", line 584, in _start_queue_management_thread
self._adjust_process_count()
File ".../python3.8/concurrent/futures/process.py", line 608, in _adjust_process_count
p.start()
File ".../python3.8/multiprocessing/process.py", line 118, in start
assert not _current_process._config.get('daemon'), \
AssertionError: daemonic processes are not allowed to have children
OK, I re-installed
cluster_tools
andmobie
through pip from the git repos but that only checks for existence and does not update dependencies.
Don't reinstall unrelated stuff through pip, but update z5py
properly via conda:
conda install "z5py>=2.0.15"
... ...cluster_tools/copy_volume/copy_volume.py", line 162, in run_impl self.submit_jobs(n_jobs, self.prefix) File ".../cluster_tools/cluster_tasks.py", line 567, in submit_jobs tasks = [pp.submit(self._submit, job_id, job_prefix) for job_id in range(n_jobs)] File ".../cluster_tools/cluster_tasks.py", line 567, in <listcomp> tasks = [pp.submit(self._submit, job_id, job_prefix) for job_id in range(n_jobs)] File ".../python3.8/concurrent/futures/process.py", line 645, in submit self._start_queue_management_thread() File ".../python3.8/concurrent/futures/process.py", line 584, in _start_queue_management_thread self._adjust_process_count() File ".../python3.8/concurrent/futures/process.py", line 608, in _adjust_process_count p.start() File ".../python3.8/multiprocessing/process.py", line 118, in start assert not _current_process._config.get('daemon'), \ AssertionError: daemonic processes are not allowed to have children
I have no idea what this is, can you point me to the script you're using?
I am trying to re-initiate my conda env from the mobie
environment file. Let's see if that fixes things.
Otherwise the script is here:
https://github.com/mobie/centriole-tomo-examples/blob/main/join2bdv.py
Are you sure you have set target to slurm
here: https://github.com/mobie/centriole-tomo-examples/blob/main/join2bdv.py#L19?
Because the code-path where the error occurs should only be executed for target="local"
, and then also makes sense since you're trying to nest processes.
Yes,
I was trying to debug the run locally. With slurm
it runs.
waiting for the results ...
Huh, something weird is going on with the data. I looks like there is some chunk-wise normalization going on.
I will convert again keeping the singed int type, but I am pretty sure it wasn't in there...
that's in /g/schwab/Tobias/MoBIE2
Yes, it clearly happens during the int_to_uint
conversion.
Compare the signed data in /g/schwab/Tobias/MoBIE_signed
. (only showing the "black" zero values)
Can you please send me the exact filepaths to a file that has the issue and is converted to unsigned int and the corresponding file that is not converted?
/g/schwab/Tobias/MoBIE2/tomo/images/ome-zarr/MMRR_07_grid12_c004.ome.zarr
vs.
/g/schwab/Tobias/MoBIE_signed/tomo/images/ome-zarr/MMRR_07_grid12_c004.ome.zarr
Ok, the issue was that this runs into the wrong code path for conversion from float dtypes to uint8
. And this conversion is not really correct in any case.
The behaviour for int_to_uint
will be fixed by #36. General fix for the uint8
conversion will come later; I will create an issue to keep track of this.
This issue is fixed now on master
and I already made a new release, which should be on conda forge after Easter.
And I made a follow-up issue for the incorrect uint8 conversion for float data: https://github.com/constantinpape/cluster_tools/issues/37
Hi,
when converting to MoBIE, I run into:
I have no idea where that
dimension_separator
comes from it seems to be notNone
otherwise https://github.com/constantinpape/cluster_tools/blob/4d25373bcad32b9eeee48bcfd5c65fa773a31e1c/cluster_tools/copy_volume/copy_volume.py#L137 should filter it. It does not appear in any of theluigi
config files.Also, like in https://github.com/mobie/mobie-utils-python/pull/57 I noticed the missing
str()
here: https://github.com/constantinpape/cluster_tools/blob/4d25373bcad32b9eeee48bcfd5c65fa773a31e1c/cluster_tools/copy_volume/copy_volume.py#L128