CLI Cheatsheet

Preflight Checks

Use substra --version to ensure your installation is working and compatible with the server according to the compatibility table.

Note: the --force-reinstall option might help if you need to install a new version of the package: pip install substra==<VERSION> --force-reinstall

You can then test that your substra CLI is able to reach the substra network with:

curl <DOMAIN_URL>/readiness

# Demo example
curl substra-backend.node-1.com/readiness

Note: If you do not see the “OK” message, we might need to check your configuration, for example the /etc/hosts file.

Substra config

You can setup a profile, this is useful if you work with several configurations, but not required, so you can omit the --profile <PROFILE_NAME> option if you don’t use a specific profile.

# Usage
substra config [OPTIONS] <URL>

# With a profile
substra config --profile <PROFILE_NAME> <URL>

# Demo example
substra config --profile node-1 http://substra-backend.node-1.com

Note: the profile name is a local resource located in ~/.substra, this means that you can use any name you want.

Substra login

# Usage
substra login [OPTIONS]

# You can simply use
substra login

# Or provide the required parameter like this
substra login --profile <PROFILE_NAME> --username <USERNAME> --password <PASSWORD>

# Demo example
substra login --profile node-1 --username node-1 --password 'p@$swr0d44'

There are also a bunch of useful options like --log-level DEBUG or --verbose.

List & Get assets

Assets you can list:

  • node: nodes in substra network

  • dataset: registered datasets

  • objective: Machine Learning question

  • algo: registered algorithm

  • compute_plan: the “blueprint” of tasks to be executed accross the substra network

substra list node # with the profile: substra list node --profile node-1
substra list dataset
substra list objective
substra list algo
substra list complute_plan

substra list traintuple
substra list testtupuble

Theses commands will return the keys of the assets that you will then need to use commands like get, describe or download.

You will then have to use:

substra get traintuple <KEY>
substra get testtuple <KEY>
substra get compute_plan <KEY>
substra get algo <KEY>

For example, you will get more information about this testuple, like its STATUS, RANK, PERF or associated PERMISSIONS:

substra get testtuple d8238915246f02c3b0f8cd4aa0a9546b31daa72477f98fddb5b2137a353d0c1c
KEY                         d8238915246f02c3b0f8cd4aa0a9546b31daa72477f98fddb5b2137a353d0c1c
TRAINTUPLE KEY              28d4c395daac2aa80f46792cbeb4572135cd0aa06042b47f400de65271eacc2f
TRAINTUPLE TYPE             traintuple
ALGO KEY                    7e039d490df04ed3f613191d3af004a1147f22b1f28b942f11a6c830caeb9e9b
ALGO NAME                   Titanic: Random Forest
OBJECTIVE KEY               1158d2f5c0cf9f80155704ca0faa28823b145b42ebdba2ca38bd726a1377e1cb
CERTIFIED                   True
STATUS                      doing
PERF                        0
DATASET KEY                 c1d9b42d9538f825109c38c4f599d47391347d198a0770331ba790b7ebcfaa40
TEST DATA SAMPLE KEYS       2 keys
RANK                        0
TAG
COMPUTE PLAN ID
LOG
CREATOR                     MyOrg1MSP
WORKER                      MyOrg1MSP
PERMISSIONS                 Processable by its owner only

Register assets

Dataset

substra add dataset [OPTIONS] PATH
substra add dataset --help

Data samples

substra add data_sample [OPTIONS] PATH
substra add data_sample --help

You can use the same pattern for registering several kind of assets: objective, compute_plan, algo, testtuple, traintuple, etc.

Tips

You can use --output formatting options like -o yaml or -o json; pretty is used by default.

Help

Remember, you can use anytime the substra CLI --help command which will provide you with a lot of resources for each asset and method!

substra --help
substra <command> --help