A critical vulnerability in the Java library Log4j was reported (see https://www.govcert.ch/blog/zero-day-exploit-targeting-popular-java-library-log4j/ for more details).

The Big Data Link Server is currently affected by this vulnerability. If you use Big Data Link, you should replace line 47 of the start-up script 'servers/big_data_link_server/bin/big_data_link_server' with:


In case of questions, please contact our support desk: openBIS Support openbis-support@id.ethz.ch

With Big Data Link it is possible to track files in openBIS without storing the files themselves into openBIS. Only the metadata is stored in openBIS.


oBIS is a command line tool which interacts with files and openBIS. It provides functionality to track file changes. This helps with data-provenance tracking. New versions of files or new files can be registered in openBIS, creating a history of data sets.

oBIS is based on pyBIS, git and git-annex. Before installing obis make sure that git and git-annex are available. 

oBIS (and pyBIS) can be downloaded from the pypi repository, from the openBIS git repository (git@sissource.ethz.ch:sispub/openbis.git) or from a built release (openBIS Download Page, Clients and APIs).

Downloading and installing from pypi is probably the easiest:

$ pip3 install obis
$ pip3 install pybis

To install oBIS from the git repository:

$ git clone git@sissource.ethz.ch:sispub/openbis.git
$ cd openbis
$ pip3 install obis/src/python
$ pip3 install obis/src/pybis

The installation process of obis wants to add manual pages to the system. If this leads to an 'access denied' error comment out the line data_files=data_files, in obis/src/python/setup.py (that is, add a hash symbol '#' in front of this line).

Typing 'obis' list all commands. Typing 'obis <command> --help' shows options and parameters of the specified command.

With 'get' you retrieve one or more settings. If the 'key' is omitted, you retrieve all settings of the 'type'.

obis [type] [options] get [key]

With 'set' you set one or more settings.

obis [type] [options] set [key1]=[value1], [key2]=[value2], ...

With 'clear' you unset one or more settings.

obis [type] [options] clear [key1]

With the type 'settings' you can get all settings at once.

obis settings [options] get

The option '-g' can be used to interact with the global settings. The global settings are stored in ~/.obis and are copied to an obis repository when that is created.

Following settings exist:

configallow_only_httpsDefault is true. If false, http can be used to connect to openBIS.
configfileservice_urlURL for downloading files. See DownloadHandler / FileInfoHandler services.
configgit_annex_backendGit annex backend to be used to calculate file hashes. Supported backends are SHA256E (default), MD5 and WORM.
configgit_annex_hash_as_checksumDefault is true. If false, a CRC32 checksum will be calculated for openBIS. Otherwise, the hash calculated by git-annex will be used.
confighostnameHostname to be used when cloning / moving a data set to connect to the machine where the original copy is located.
configopenbis_urlURL for connecting to openBIS (only protocol://host:port, without a path).
configobis_metadata_folderAbsolute path to the folder which obis will use to store its metadata. If not set, the metadata will be stored in the same location as the data. This setting can be useful when dealing with read-only access to the data. The clone and move commands will not work when this is set.
configuserUser for connecting to openBIS.
data_settypeData set type of data sets created by obis.
data_setpropertiesData set properties of data sets created by obis.
objectidIdentifier of the object the created data set is attached to. Use either this or the collection id.
collectionidIdentifier of the collection the created data set is attached to. Use either this or the object id.
repositorydata_set_idThis is set by obis. Is is the id of the most recent data set created by obis and will be used as the parent of the next one.
repositoryexternal_dms_idThis is set by obis. Id of the external dms in openBIS.
repositoryidThis is set by obis. Id of the obis repository.

The settings are saved within the obis repository, in the .obis folder, as JSON files, or in ~/.obis for the global settings. They can be added / edited manually, which might be useful when it comes to integration with other tools.

    "fileservice_url": null,
    "git_annex_hash_as_checksum": true,
    "hostname": "bsse-bs-dock-5-160.ethz.ch",
    "openbis_url": "http://localhost:8888"
    "properties": {
        "K1": "v1",
        "K2": "v2"
    "type": "UNKNOWN"
obis init [folder]

If a folder is given, obis will initialize that folder as an obis repository. If not, it will use the current folder.

obis init_analysis [options] [folder]

With init_analysis, a repository can be created which is derived from a parent repository. If it is called from within a repository, that will be used as a parent. If not, the parent has to be given with the '-p' option.

obis commit [options]

The 'commit' command adds files to a new data set in openBIS. If the '-m' option is not used to define a commit message, the user will be asked to provide one.

obis sync

When git commits have been done manually, the 'sync' command creates the corresponding data set in openBIS. Note that, when interacting with git directly, use the git annex commands whenever applicable, e.g. use "git annex add" instead of "git add".

obis status [folder]

This shows the status of the repository folder from which it is invoked, or the one given as a parameter. It shows file changes and whether the repository needs to be synchronized with openBIS.

obis clone [options] [data_set_id]

The 'clone' command copies a repository associated with a data set and registers the new copy in openBIS. In case there are already multiple copied of the repository, obis will ask from which copy to clone. To avoid user interaction, the copy index can be chosen with the option '-c'. With the option '-u' a user can be defined for copying the files from a remote system. By default, the file integrity is checked by calculating the checksum. This can be skipped with '-s'.

Note: This command does not work when obis_metadata_folder is set.

obis move [options] [data_set_id]

The 'move' command works the same as 'clone', except that the old repository will be removed.

Note: This command does not work when obis_metadata_folder is set.

obis download [options] [data_set_id]

The 'download' command downloads the files of a data set. Contrary to 'clone', this will not register another copy in openBIS. It is only for accessing files. This command requires the DownloadHandler / FileInfoHandler microservices to be running and the 'fileservice_url' needs to be configured.

obis addref
obis removeref

Obis repository folders can be added or removed from openBIS. This can be useful when a repository was moved or copied without using the 'move' or 'copy' commands.

# global settings to be use for all obis repositories
obis config -g set openbis_url=https://localhost:8888
obis config -g set user=admin
# create an obis repository with a file
obis init data1
cd data1
echo content >> example_file
# configure the repository
obis data_set set type=UNKNOWN
obis object set id=/DEFAULT/DEFAULT
# commit to openBIS
obis commit -m 'message'
# assuming we are in a configured obis repository
echo content >> example_file
git annex add example_file
git commit -m 'message'
obis sync
# assuming we have a repository 'data1'
obis init_analysis -p data1 analysis1
cd analysis1
obis data_set set type=UNKNOWN
obis object set id=/DEFAULT/DEFAULT
echo content >> example_file
obis commit -m 'message'

Big Data Link Services

The Big Data Link Services can be used to download files which are contained in an obis repository. The services are included in the installation folder of openBIS, under servers/big_data_link_services. For how to configure and run them, consult the README.md file.

  • No labels