ckanext-dcor-depot 0.13.8

Manages data storage for CKAN/DCOR (import, symlink, etc.)

This plugin manages how data are stored in DCOR. There are two types of files in DCOR:

1. Resources uploaded by users, imported from figshare, or imported from a data archive

2. Ancillary files that are generated upon resource creation, such as condensed DC data, preview images (see ckanext-dc_view).

This plugin implements:

• Data storage management. All resources uploaded by a user are moved to /data/users-HOSTNAME/USERNAME-ORGNAME/PK/ID/PKGNAME_RESID_RESNAME and symlinks are created in /data/ckan-HOSTNAME/resources/RES/OUR/CEID via a background job. CKAN itself will not notice this. The idea is to have a filesystem overview about the datasets of each user.

• A backround job uploads all resources to S3 after_resource_create.

• Import datasets from figshare. Existing datasets from figshare are downloaded to the /data/depots/figshare directory and, upon resource creation, symlinked there from /data/ckan-HOSTNAME/resources/RES/OUR/CEID (Note that this is an exemption of the data storage management described above). When running the following command, the “figshare-import” organization is created and the datasets listed in figshare_dois.txt are added to CKAN:

  ckan import-figshare

• Populate an internal depot from RT-DC data stored in tar archives. This is part of an effort to have automated imports of RT-DC data from other sources. The idea is to move experimental data to the DCOR server in tar archives and DCOR can then populate the internal depot with it. The location of the internal depot is /data/depots/internal/ and it follows a very specific directory structure 201X/2019-08/20/2019-08-20_1126_c083de* where the path is generated from the acquisition date, time, and part of the hash (c083de) of the original data file. According to this scheme, all files with the same path stem belong to one dataset:

  • 2019-08-20_1126_c083de.sha256sums a file containing SHA256 sums

  • 2019-08-20_1126_c083de_v1.rtdc the actual measurement

  • 2019-08-20_1126_c083de_v1_condensed.rtdc the condensed dataset

  • 2019-08-20_1126_c083de_ad1_m001_bg.png an ancillary image

  • 2019-08-20_1126_c083de_ad2_m002_bg.png another ancillary image

  • …

  ckan depotize-archive

• Import datasets from the internal depot. The previous command depotize-archive just populates the depot directory structure. To make the datasets available in CKAN, this step must be performed:

  ckan import-internal

• Upgrade datasets from the internal depot. If you find issues or have to add new versions of an .rtdc resource, simply create the _v2.rtdc files and run this command:

  ckan upgrade-internal

• CLI for symlinking datasets that have failed to symlink before:

  ckan run-jobs-dcor-depot

• CLI for appending a resource to a dataset

  ckan append-resource /path/to/file dataset_id --delete-source

Please make sure that the necessary file permissions are given in /data.

In 2023, it was decided that the huge block storage of DCOR should be replaced with an S3-compatible object store, because block storage does not scale well. This partially deprecates some of the commands above which might be removed or modified to support object storage directly.

• CLI for migrating data from block storage to an S3-compatible object storage service. For this, the following configuration keys must be specified in the ckan.ini file:

  dcor_object_store.access_key_id = ACCESS_KEY_ID
  dcor_object_store.secret_access_key = SECRET_ACCESS_KEY
  dcor_object_store.endpoint_url = S3_ENDPOINT_URL
  dcor_object_store.ssl_verify = true
  # The bucket name is by default defined by the circle ID. Resources
  # are stored in the "RES/OUR/CEID-SCHEME" in that bucket.
  dcor_object_store.bucket_name = circle-{organization_id}

  Usage:

  ckan dcor-migrate-resources-to-object-store

Installation

pip install ckanext-dcor_depot

Add this extension to the plugins and defaul_views in ckan.ini:

ckan.plugins = [...] dcor_depot
ckan.storage_path=/data/ckan-HOSTNAME
ckanext.dcor_depot.depots_path=/data/depots
ckanext.dcor_depot.users_depot_name=users-HOSTNAME

This plugin stores resources to /data:

mkdir -p /data/depots/users-$(hostname)
chown -R www-data /data/depots/users-$(hostname)

Testing

If CKAN/DCOR is installed and setup for testing, this extension can be tested with pytest:

pytest ckanext

Testing can also be done via vagrant in a virtualmachine using the dcor-test <https://app.vagrantup.com/paulmueller/boxes/dcor-test/> image. Make sure that vagrant and virtualbox are installed and run the following commands in the root of this repository:

# Setup virtual machine using `Vagrantfile`
vagrant up
# Run the tests
vagrant ssh -- sudo bash /testing/vagrant-run-tests.sh