album plugin for running a solution repeatedly and distributed
Project description
Album plugin for distributed calls
This is an early version of enhancing album with calls for batch and distributed processing.
Installation
- install Album
- Activate the album environment:
conda activate album
- Install this plugin:
pip install https://gitlab.com/album-app/plugins/album-distributed/-/archive/main/album-distributed-main.zip
Usage
Fist, install a solution - replace solution.py
with the path to your solution / solution folder or with the group:name:version
coordinates of your solution.
album install solution.py
Now you can use the plugin:
album run-distributed solution.py
The plugin does two things:
- It figures out if the input arguments match multiple tasks - in this case, it generates the different task arguments.
- It runs all matching tasks, the mode for running these tasks can be chosen.
Since the matching part can be tricky, please use the --dry-run
argument to first print a list of matched tasks:
album run-distributed solution.py --dry-run
On Windows, replace the slashes with backslashes in the examples on this page.
Please let us know if you run into issues.
Matching input arguments
To generate multiple tasks, patterns in file name arguments can be used to match multiple files.
Using patterns in a single argument
You should be able to use all glob
features when using it in a single argument. Here are some examples:
In the following scenarios solution.py
has an argument called input_data
.
Match all .tif
files in the current folder:
album run-distributed solution.py --input_data *.tif
Match all .tif
files in a specific folder where the file name starts with input
:
album run-distributed solution.py --input_data /data/input*.tif
Match all .tif
files recursively, starting from the current folder:
album run-distributed solution.py --input_data **/*.tif
Using patterns in multiple arguments
When using patterns in multiple arguments, this plugin will try to figure out the corresponding argument values based on which of the patterns match with existing files.
This is likely to fail in a bunch of situations - please use the --dry-run
argument to test if the matched tasks correspond with your expectation.
In the following scenarios solution.py
has two arguments called input_data
and output_data
.
Use all .tif
files in the current folder and append _out
to the file name for the output argument.
album run-distributed solution.py --input_data *.tif --output_data *_out.tif
Do the same thing recursively:
album run-distributed solution.py --input_data **/*.tif --output_data **/*_out.tif
Let the output argument values live in a different folder:
album run-distributed solution.py --input_data *.tif --output_data output/*.tif
Since Album does not yet distinguish between input and output arguments, be aware that if the output_data
argument in these scenarios matches existing files, the plugin will also try to generate corresponding input_file
values. We will work on improving this.
Modes
You can set the mode by using the --mode
argument:
album run-distributed solution.py --mode queue
By default, the plugin will use the basic
mode.
Basic
In this mode, all tasks will be performed one after each other. The console output of each task will be printed.
Queue
In this mode, a set of thread workers will be created to process tasks in parallel. The console output of each task will not be printed.
You can control how many threads should be created with the --threads
argument:
album run-distributed solution.py --mode queue --threads 16
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
File details
Details for the file album-distributed-0.1.0.tar.gz
.
File metadata
- Download URL: album-distributed-0.1.0.tar.gz
- Upload date:
- Size: 10.0 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.7.1 importlib_metadata/4.11.3 pkginfo/1.8.2 requests/2.27.1 requests-toolbelt/0.9.1 tqdm/4.64.0 CPython/3.8.13
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | e70645d7d5d08ac60971bb11c2baf6b7127cd8221927054e7d9cb6c34f461010 |
|
MD5 | 5df1dfd4103c3c3638da6bba46ff0fb7 |
|
BLAKE2b-256 | 9f517c4a6b33e644eb133d08f8aefcac781ba335d8f220dee0e4b87e8609bbb6 |