Skip to content



This section walks you through an example of using Pachyderm's experimental SAML support. We'll describe:

  1. Authenticating via a SAML ID Provider
  2. Authenticating in the CLI
  3. Authorizing a user or group to access data


Follow the instructions in Configure SAML to enable auth in a Pachyderm cluster and connect it to a SAML ID provider. Then, we'll authenticate as a cluster admin in one console and set up our open CV demo.

In the CLI, that would look like:

(admin)$ pachctl auth use-auth-token

System response:

Please paste your Pachyderm auth token:
<auth token>
(admin)$ pachctl auth whoami

System response:

You are "robot:admin"
You are an administrator of this Pachyderm cluster

Create a repository, pipelines, and put files into the repository.

(admin)$ pachctl create repo images
(admin)$ pachctl create pipeline -f examples/opencv/edges.json
(admin)$ pachctl create pipeline -f examples/opencv/montage.json
(admin)$ pachctl put file images@master -i examples/opencv/images.txt
(admin)$ pachctl put file images@master -i examples/opencv/images2.txt

View the list of existing repositories:

(admin)$ pachctl list repo

System response:

montage 2 minutes ago 1.653MiB      OWNER
edges   2 minutes ago 133.6KiB      OWNER
images  2 minutes ago 238.3KiB      OWNER

View the list of jobs:

(admin)$ pachctl list job

System response:

ID                               OUTPUT COMMIT                            STARTED       DURATION  RESTART PROGRESS  DL       UL       STATE
023a478b16e849b4996c19632fee6782 montage/e3dd7e9cacc5450c92e0e62ab844bd26 2 minutes ago 8 seconds 0       1 + 0 / 1 371.9KiB 1.283MiB success
fe8b409e0db54f96bbb757d4d0679186 edges/9cc634a63f794a14a78e931bea47fa73   2 minutes ago 5 seconds 0       2 + 1 / 3 181.1KiB 111.4KiB success
152cb8a0b0854d44affb4bf4bd57228f montage/82a49260595246fe8f6a7d381e092650 2 minutes ago 5 seconds 0       1 + 0 / 1 79.49KiB 378.6KiB success
86e6eb4ae1e74745b993c2e47eba05e9 edges/ee7ebdddd31d46d1af10cee25f17870b   2 minutes ago 4 seconds 0       1 + 0 / 1 57.27KiB 22.22KiB success

Authenticating via a SAML ID Provider (in the dashboard)

Before authenticating, navigating to the dash will yield a blank screen:

Blocked-out dash

Even through the dash suggests logging in via GitHub, we will log in using a SAML IdP (which has hopefully already been configured). To see your Pachyderm DAG, navigate to your SAML ID provider and sign in to your Pachyderm cluster there (currently Pachyderm only supports IdP-initiate SAML authentication).

SSO image

Once you've authenticated, you'll be redirected to the Pachyderm dash (the redirect URL is configured in the Pachyderm auth system). You'll be given the opportunity to generate a one-time password (OTP), though you can always do this later from the settings panel.

Dash logged in

After closing the OTP panel, you'll be able to see the Pachyderm DAG, but you may not have access to any of the repos inside (a repo that you cannot read is indicated by a lock symbol):

Dash with locked repos

Authenticating in the CLI

After authenticating in the dash, you'll be given the opportunity to generate a one-time password (OTP) and sign in on the CLI. You can also generate an OTP from the settings panel:

OTP Image

(user)$ pachctl auth login --code auth_code:73db4686e3e142508fa74aae920cc58b
(user)$ pachctl auth whoami

System response:

You are ""
session expires: 14 Sep 18 20:55 PDT

Note that this session expires after 8 hours. The duration of sessions is configurable in the Pachyderm auth config, but it's important that they be relatively short, as SAML group memberships are only updated when users sign in. If a user is removed from a group, they'll still be able to access the group's resources until their session expires.

Authorizing a user or group to access data

First, we'll give the example of an admin granting a user access. This can be accomplished on the CLI like so:

(admin)$ pachctl auth set reader images

Now, the images repo is no longer locked when that user views the DAG:

Unlocked images repo image

At this point, you can click on the images repo and preview data inside:

Unlocked images repo image

Likewise, you can grant access to repos via groups. You'll need a SAML ID provider that supports group attributes, and you'll need to put the name of that attribute in the Pachyderm auth config. Here, we'll grant access to the Everyone group:

(admin)$ pachctl auth set group/saml:Everyone owner edges

Now, the edges repo is also not locked:

Unlocked edges repo

Also, becase has OWNER provileges in the edges repo (via the Everyone group), the ACL for edges can be edited. will use OWNER privileges gained via the Everyone group to add (the user principal) directly to that ACL:

Adding user to ACL image

this change is reflected in the CLI as well:

(admin)$ pachctl auth get edges

System response:

pipeline:edges: WRITER
pipeline:montage: READER
group/saml:Everyone: OWNER READER
robot:admin: OWNER


This is just an example of Pachyderm's auth system, meant to illustrate the general nature of available features. Hopefully, it clarifies whether Pachyderm can meet your requirements.

Last update: January 5, 2021
Does this page need fixing? Edit me on GitHub