Luke, it worked!
I changed the CUSTOM_COMMANDS on the juliaset example to:
"CUSTOM_COMMANDS = [
['apt-get', 'update'],
['apt-get', 'install', 'ghostscript', '-y']]"
Thank you very much!!
On Thu, Jan 9, 2020 at 3:03 PM Luke Cwik <[email protected]> wrote:
> Andre, add the required installation commands (e.g. the apt-get install
> commands) for the non-Python dependencies to the list of CUSTOM_COMMANDS in
> your setup.py file. See the Juliaset setup.py [1] for an example. Note: You
> must make sure that these commands are runnable on the remote worker (e.g.
> if you use apt-get, the remote worker needs apt-get support).
>
> See the website about managing Python dependencies [2] for more details.
>
> 1:
> https://github.com/apache/beam/blob/master/sdks/python/apache_beam/examples/complete/juliaset/setup.py
> 2:
> https://beam.apache.org/documentation/sdks/python-pipeline-dependencies/#nonpython
>
> On Thu, Jan 9, 2020 at 7:36 AM Leonardo Campos <
> [email protected]> wrote:
>
>> Hi, Andre,
>>
>> On a very different topic, I was trying to find a way to change the JVM
>> default encoding and could not find a way to do so.
>> In this sense, it would also be of my interest to be able to influence
>> the image used by the workers.
>>
>> Sorry for having no help,
>> Leonardo Campos
>> On 1/9/20 3:48 PM, André Rocha Silva wrote:
>>
>> Hi all!
>>
>> I am trying to use imagemagick on Dataflow [Apache Beam Python 3.7 SDK
>> 2.17.0], but I am facing a problem. The function works properly local, but
>> when I use it in Dataflow I receive this message:
>> File "/usr/local/lib/python3.7/site-packages/wand/image.py", line 7888,
>> in read raise WandRuntimeError(msg) wand.exceptions.WandRuntimeError:
>> MagickReadImage returns false, but did raise ImageMagick exception. This
>> can occurs when a delegate is missing, or returns EXIT_SUCCESS without
>> generating a raster.
>> As far as I researched, it may be because ghostscript is not installed.
>> Then I would like to know:
>>
>> 1) Is there a way to change the image that Dataflow loads for its
>> workers? If so, I could install the programs I need for the job
>>
>> 2) Can I ask Dataflow to install a program for all workers? It is not
>> feasible to run a " os.system(apt-get install ghostscript)" for every
>> single element, isn't it?
>>
>> If someone has faced this problem and solved in another way, I am totally
>> open to suggestions.
>>
>> Thank you
>> André Rocha Silva
>>
>>
