DTaaS Command Line Interface
Command-line tool for the
INTO-CPS-Association
Digital Twin as a Service platform. Use it to generate deployment projects,
manage users, and operate a running DTaaS instance. All from a single dtaas
entry point.
Quick Start
From a clean machine to a running DTaaS deployment in seven steps.
# 1. Create and activate a virtual environment (recommended)
python -m venv .venv && source .venv/bin/activate # Linux / macOS
# python -m venv .venv && .venv\Scripts\activate # Windows
# 2. Install the package
pip install dtaas
# 3. Generate dtaas.toml + a sample users.csv to fill in
dtaas config generate
# 4. Open dtaas.toml and fill in your server DNS, paths, and credentials
# (see Configuration Reference below for all fields)
# 5. Validate the configuration fix any reported errors before continuing
dtaas config validate
# 6. Generate deployment files for your chosen scenario
dtaas deployment generate --type secure-server
# 7. Bring the deployment up
dtaas platform install
# Tear it down when done
dtaas platform uninstall
Deployment type cheat-sheet
--typeWhen to use localhostLocal dev / demo only insecure-serverMulti-user HTTP demo: not internet-facing secure-serverMulti-user HTTPS: production-ready secure-server-gitlabHTTPS + bundled GitLab: production-ready workspace-localhostWorkspace + Dex on localhost workspace-secure-serverWorkspace + Keycloak: production-ready
Table of Contents
- DTaaS Command Line Interface
Installation
Installation inside a virtual environment is strongly recommended to avoid conflicts with system-wide packages.
python -m venv .venv
source .venv/bin/activate # Linux / macOS
# .venv\Scripts\activate # Windows
pip install dtaas
Verify the install:
dtaas --help
dtaas --version # which build is installed, the first thing to pin down
# when a command behaves unexpectedly
Commands
config
Manage dtaas.toml independently of the rest of the project. This is the
first step in the setup workflow generate a template, fill it in, then
validate before running any other command.
Generate a fresh template
dtaas config generate
This writes dtaas.toml and a sample users.csv (bulk input for
dtaas user add --file) into the target directory.
Validate an existing file
dtaas config validate
validate reads dtaas.toml (from --output-dir first, then the current
directory) and reports all problems at once:
| Field | Rule |
|---|---|
git-repo |
Must be an http(s) URL |
[common].server-dns |
Must be localhost, an IP, or a fully qualified hostname |
[common].path |
Must be an absolute path to an existing directory |
[common.security].certs-src |
When present, must be an absolute path to an existing directory |
[common.resources].set_limits |
false turns the limits off; any other value, or no key, keeps them on |
[common.resources].cpus |
Positive number (e.g. 4 or 0.5) |
[common.resources].pids_limit |
Integer |
[common.resources].mem_limit, shm_size |
Byte size with required unit (e.g. 4G, 512m) |
[[users]] |
When present, must be an array of tables; usernames must be unique |
[[users]].username |
Required, valid username |
[[users]].email |
Required, valid RFC 5321/5322 address (no DNS lookup) |
[[users]].groups |
When present, must be a list of strings |
[[users]].load_balance |
When present, must be true or false |
[[users]].password |
When present, must be a string (the starting user's initial GitLab password, used when user add names them) |
[gitlab].provision |
When present, true or false (default false) |
[gitlab].api_url |
Must be an http(s) URL; required when provision is true |
[gitlab].pat |
When present, must be non-empty (remove the key to use DTAAS_GITLAB_PAT) |
[gitlab].ssl_verify |
When present, true, false, or a CA-bundle path string |
| Deployment-section URLs | When present, must be http(s) URLs |
Deployment-section default-user |
When present, must be a valid username |
Deployment-section URLs are react-app-oauth-url, oauth-url,
auth-authority, and keycloak-issuer-url across [frontend],
[localhost], [insecure-server], [secure-server],
[secure-server-gitlab], [workspace-localhost], and
[workspace-secure-server]; each is checked only when its section is
present. config validate checks every section in the file, while
platform update --config checks only the installed type's section, plus
[frontend].
path and certs-src are checked against the local filesystem, run
validate on the deployment host. If a check fails with a permission error
(e.g. certs-src points under /etc/letsencrypt, which is root-owned), the
CLI says so and prints the elevated form to re-run:
sudo -E env PATH="$PATH" dtaas <command>.
The [common.resources] limit fields (cpus, pids_limit, mem_limit,
shm_size) are required only when set_limits is true (the default). With
set_limits = false they are optional and ignored; any value still present is
validated.
Options:
| Option | Default | Description |
|---|---|---|
--output-dir PATH |
. |
For generate: target directory (created if missing). For validate: search location |
--force |
off | (generate only) Overwrite an existing dtaas.toml |
deployment generate
Copies the full project structure for a specific deployment scenario
(docker-compose.yml, config examples, and supporting files) into a target
directory ready to be customised. It also writes the user workspace
templates that user add builds new workspaces from: users.server.yml,
users.server.secure.yml, users.resources.yml, and the files/template/
skeleton.
dtaas deployment generate --type <name>
Options:
| Option | Default | Description |
|---|---|---|
--type NAME |
(required) | Deployment scenario (see table below) |
--output-dir PATH |
. |
Target directory (created if missing) |
--force |
off | Overwrite files that already exist |
Available types
--type |
Deployment scenario | Support level |
|---|---|---|
localhost |
Single-machine Docker deployment | dev/demo only |
insecure-server |
Multi-user HTTP server | insecure/demo only |
secure-server |
Multi-user HTTPS/TLS server | production-supported |
secure-server-gitlab |
HTTPS/TLS + integrated GitLab | production-supported |
workspace-localhost |
Workspace service with Dex on localhost | dev/demo only |
workspace-secure-server |
Workspace service with Keycloak | production-supported |
Warning Types marked dev/demo only or insecure/demo only run over plain HTTP with default or static credentials. Do not expose them to the internet or shared networks. Production-supported types still require manual hardening steps documented in the
README.mdand configuration notes (CONFIG.md, orCONFIGURATION.mdforworkspace-secure-server) shipped with each generated project.
Examples:
# Localhost demo in the current directory
dtaas deployment generate --type localhost
# Production HTTPS server in a subdirectory
dtaas deployment generate --type secure-server --output-dir ./my-server
# Regenerate, overwriting existing files
dtaas deployment generate --type insecure-server --output-dir ./demo --force
Configuration substitution
When dtaas.toml is present, deployment generate reads
deployment-specific values from it and substitutes them into the generated
files automatically.
The CLI searches --output-dir first, then falls back to the current working
directory. Each --type reads from its matching top-level section in
dtaas.toml. Values are written into dotenv files (config/.env,
config/conf.server) and the React client config (config/client.js).
The [frontend] section supplies REACT_APP_CLIENT_ID and
REACT_APP_AUTH_AUTHORITY for the DTaaS web client, these are a separate
OAuth application from the traefik-forward-auth credentials configured in
[insecure-server] / [secure-server]. The [common] and [[users]]
sections are substituted across all types.
If dtaas.toml is not found, a note is printed and generated files keep
their default placeholder values.
TLS certificate placement
For the TLS types (secure-server, secure-server-gitlab,
workspace-secure-server), deployment generate also populates the
certs/ directory in the output. It reads [common.security].certs-src from
dtaas.toml and copies the latest fullchain.pem and privkey.pem there.
Tip: verify the Docker image tag
users.server.ymlandusers.server.secure.ymlcontain a pinned workspace image tag (e.g.intocps/workspace:main-56c6f68). Check Docker Hub and update the tag to a current, stable version before deploying.
platform install
Brings a generated deployment up with a single command.
dtaas platform install
Internally runs docker compose up -d against the docker-compose.yml in
the installation directory. Before starting, it ensures per-user workspace
directories for every [[users]] record exist, recreating each from
files/template/ if missing and sets ownership to 1000:100.
Options:
| Option | Default | Description |
|---|---|---|
--output-dir PATH |
. |
Installation directory containing the generated deployment |
The CLI looks for dtaas.toml in --output-dir first, then the current
working directory, so a single top-level dtaas.toml can serve a deployment
generated into a subdirectory:
dtaas platform install --output-dir ./insecure
The command fails with a clear error if docker-compose.yml is missing,
dtaas.toml cannot be found, or the Docker daemon is unreachable.
platform uninstall
Tears the deployment down, stopping and removing containers and networks.
dtaas platform uninstall
User containers added with user add run as a separate Compose project;
they are torn down first so they do not hold the shared network open.
Per-user workspace files are preserved by default.
To also delete the generated per-user workspace directories and the
CLI-owned dtaas.users.registry.json / .dtaas.state.json:
dtaas platform uninstall --remove-user-files
This is destructive, so the command prompts for confirmation. Skip the prompt
in non-interactive scripts with --yes:
dtaas platform uninstall --remove-user-files --yes
Options:
| Option | Default | Description |
|---|---|---|
--output-dir PATH |
. |
Installation directory |
--remove-user-files |
off | Also delete per-user workspace dirs + registry/state files |
--yes / -y |
off | Skip the confirmation prompt for --remove-user-files |
--remove-user-filesremoves only the per-user directories inside<output-dir>/files/, preservingfiles/common/andfiles/template/so a laterplatform installcan recreate user directories. It refuses to follow a symlinkedfiles/. Double-check--output-dirbefore using this flag.
Lifecycle operations
Operational controls for an already-installed deployment: observe it with
status, and suspend or resume it with stop/start and pause/resume.
None of these remove containers or networks, that is uninstall's job. They
exit 0 on success (including the idempotent "nothing installed" case) and
non-zero on failure, so they are safe to call from CI/ops scripts.
platform stop/start/pause/resume act on the core services only —
they never touch per-user containers. Suspend or resume individual additional
users with dtaas user stop/pause/resume
instead. platform status, being read-only, still reports the whole
installation (core services and user containers).
Lifecycle command matrix:
| Command | docker compose verb |
Effect | Containers kept? | Reverse with |
|---|---|---|---|---|
platform install |
up -d |
Create and start every service | n/a | platform stop / platform uninstall |
platform status |
ps (read-only) |
Report per-service state; no change | n/a | n/a |
platform stop |
stop |
Terminate the processes, keep the containers | yes | platform start |
platform start |
start |
Start previously stopped containers | yes | platform stop |
platform pause |
pause |
Freeze the processes (memory preserved) | yes | platform resume |
platform resume |
unpause |
Thaw previously paused processes | yes | n/a |
platform uninstall |
down |
Stop and remove containers and networks | no | platform install |
stopvspause.stopsendsSIGTERM/SIGKILL: processes end, and a restart re-runs them from scratch (reverse withplatform start).pauseuses the kernel cgroup freezer: processes are suspended in place with their memory intact and resume instantly (reverse withplatform resume), but a paused container still holds its resources. Usestopto free CPU; usepausefor a brief, instantly reversible suspension.pauseexpects running containers and will error if the deployment is already stopped.
platform status
Reports the state of every service, for both the deployment and user workloads.
dtaas platform status
PROJECT SERVICE STATE
deployment traefik running
deployment client running
deployment gitlab not created
users alice paused
State values are running, paused, stopped (a terminated container, what
Docker calls exited), restarting, or not created (a service defined in
docker-compose.yml that has no container yet).
For automation, --json emits the same records as machine-readable JSON:
dtaas platform status --json
[
{"project": "deployment", "service": "traefik", "state": "running"}
]
Options:
| Option | Default | Description |
|---|---|---|
--output-dir PATH |
. |
Installation directory |
--json |
off | Emit machine-readable JSON instead of the table |
platform stop / platform start
stop stops the core services with docker compose stop; start brings the
stopped containers back with docker compose start. Containers and networks
are kept, so stop is not an uninstall.
dtaas platform stop
dtaas platform start
After stop, the command notes how many per-user containers are still
running and suggests dtaas user stop --all. Both commands print
There is no existing DTaaS / Workspace installation and exit 0 when
nothing is installed (no containers in any state), so they are safe to call
repeatedly.
Options:
| Option | Default | Description |
|---|---|---|
--output-dir PATH |
. |
Installation directory |
platform pause / platform resume
pause freezes the running core service containers in place with
docker compose pause; resume thaws them with docker compose unpause.
Memory is preserved and resume is near-instant.
dtaas platform pause
dtaas platform resume
Like stop, pause notes any per-user containers still running. Both
report the absent-installation case and exit 0 when nothing is installed
(no containers in any state).
Options:
| Option | Default | Description |
|---|---|---|
--output-dir PATH |
. |
Installation directory |
platform update --certs
Rotates TLS certificates of a running deployment in place: no project regeneration or manual file copying required.
dtaas platform update --certs
The command reads [common.security].certs-src from dtaas.toml, then:
- Validates the new certificate pair (parseable, private key matches cert, no expired intermediates).
- Stops the
traefikservice to release open file handles. - Swaps the validated files into
<output-dir>/certs/, backing up the live pair and restoring it on any failure. - Restricts the private key to
0600(POSIX; prints a warning on Windows). - Restarts
traefikand waits for it to come back up.
If validation fails, live certificates are left untouched. The command is safe to run repeatedly.
Options:
| Option | Default | Description |
|---|---|---|
--certs |
(required) | Refresh the deployment's TLS certificates |
--output-dir PATH |
. |
Installation directory |
platform update --config
Re-applies the values in dtaas.toml to an already-installed deployment's
service config files without regenerating the project.
dtaas platform update --config
Treats dtaas.toml as the single source of truth, re-runs the same
substitution as deployment generate, and if anything changed, recreates all
deployment services with docker compose up -d --force-recreate. The
deployment type is auto-detected from docker-compose.yml.
# Preview changes without writing or restarting
dtaas platform update --config --dry-run
# Apply changes and restart
dtaas platform update --config
# Update a deployment in a subdirectory
dtaas platform update --config --output-dir ./my-server
--config validates dtaas.toml before making any changes and refuses to
apply if problems are found. It is idempotent a second run with no
dtaas.toml changes reports No configuration changes and restarts nothing.
Options:
| Option | Default | Description |
|---|---|---|
--config |
(required) | Re-apply dtaas.toml to installed service config files |
--dry-run |
off | Report what would change without writing or restarting |
--output-dir PATH |
. |
Installation directory |
--certsand--configmay be combined in a single invocation; at least one of them is required.
user add
Provisions users on a running DTaaS instance. Additional users are recorded in
the CLI-owned dtaas.users.registry.json
(see User files), not in dtaas.toml.
Options:
| Option | Default | Description |
|---|---|---|
USERNAME |
— | Add one user (requires --email, except for a dtaas.toml starting user) |
--file PATH / -f |
— | Bulk-add users from a CSV |
--email TEXT |
— | Email for USERNAME (enables forward-auth routing); read from [[users]] for a starting user, and another address given for one is reported and not applied |
--group TEXT |
additional |
Group tag for USERNAME; repeat the flag for multiple groups, e.g. --group dtaas --group testers |
--load-balance / --no-load-balance |
on | Mark USERNAME for load balancing |
--password TEXT |
— | GitLab password for USERNAME; only used when GitLab provisioning is enabled (see below). It wins over [[users]].password in dtaas.toml. Visible in shell history and the process list, prefer the users.csv password column (chmod 600 it) or the interactive prompt for non-interactive/scripted use |
Add a single user:
dtaas user add --email alice@intocps.org --group dtaas --load-balance alice
Click accepts
--group, and--load-balancein any position relative toUSERNAMEthe form above is the recommended convention, matchinguseradd [options] LOGIN.
--group is repeatable, not comma-separated pass it once per group to add a
user to multiple groups:
dtaas user add --email alice@intocps.org --group dtaas --group testers alice
Or bulk-add from a CSV:
dtaas user add --file users.csv
dtaas config generate writes a sample users.csv next to dtaas.toml:
username,email,groups,load_balance,password
alice,alice@intocps.org,additional,true,
bob,bob@intocps.org,additional;beta-testers;load-balanced-tier1,false,
groups is a ;-separated list and load_balance is true/false. The
password column is used only with GitLab provisioning (see below). Both
forms merge into the registry (never hand-edited) with desired_status set to
running, then only the newly-added users are started: already-running
users are left untouched, so adding one user never recreates the rest. A
username already in the registry is reported ('alice' already exists, skipping) and never added twice or overwritten. A username declared in
dtaas.toml's [[users]] is reported as a starting user, whose workspace
dtaas.toml manages; with GitLab provisioning on, naming one creates its
GitLab account (see below).
A USERNAME or --file is required (not both): a bare dtaas user add
with neither is rejected rather than silently reprovisioning the whole
registry. A run that added nobody (every name was already registered, or was
a starting user) closes with No new users added. and still exits 0, so
re-running the same CSV stays idempotent. What the deployment supports is
checked before anything is registered, so a run that cannot provision (a
localhost installation, or a missing per-user template) leaves no users
registered behind it. To (re)provision every registry user at once (for
example after compose.users.yml was lost), use dtaas config reconcile --fix instead.
GitLab provisioning (optional)
With [gitlab].provision = true, user add also creates a GitLab account
for each user it adds, and issues that account a Personal Access Token
(PAT). The containers are provisioned first, and a GitLab failure never
undoes them.
Setup, once per installation. Turn provisioning on in dtaas.toml:
[gitlab]
provision = true
api_url = "https://gitlab.example.com"
Then export an admin token. It must belong to a GitLab administrator and
carry the api scope, plus admin_mode when GitLab Admin Mode is on:
export DTAAS_GITLAB_PAT="glpat-xxxxxxxxxxxxxxxxxxxx"
[gitlab].pat in dtaas.toml is the fallback when the variable is unset.
Example 1: add one user. Without --password, the CLI asks for the
initial GitLab password with hidden input:
$ dtaas user add --email alice@intocps.org alice
GitLab password for 'alice':
Repeat for confirmation:
GitLab account and token created for 'alice'.
The token is in gitlab_user_tokens.json.
Users added successfully
Example 2: add users from a CSV. Fill in the password column of the
users.csv that dtaas config generate writes, and keep the file private:
username,email,groups,load_balance,password
alice,alice@intocps.org,additional,true,Kq7v-Rm3x-Zt9w
bob,bob@intocps.org,additional;beta-testers,false,Wd4p-Hs8n-Qc2y
$ chmod 600 users.csv
$ dtaas user add --file users.csv
GitLab account and token created for 'alice'.
The token is in gitlab_user_tokens.json.
GitLab account and token created for 'bob'.
The token is in gitlab_user_tokens.json.
Users added successfully
A row with an empty password cell refuses the whole run before anything
is changed, naming the users concerned.
Example 3: a starting user. Naming a [[users]] user of dtaas.toml
creates only their GitLab account and token. Their workspace and email come
from dtaas.toml, and the password from [[users]].password (or the
prompt):
$ dtaas user add username1
'username1' is a dtaas.toml starting user.
Its workspace is managed there, not by 'user add'.
GitLab account and token created for 'username1'.
The token is in gitlab_user_tokens.json.
No new users added.
Example 4: retry after a failure. Suppose GitLab had refused bob's
password in example 2: alice got her account, bob got only his container,
and the command exited non-zero. Fix the password in users.csv and run the
same command again. Only the missing work is done, and alice is not issued
a second token:
$ dtaas user add --file users.csv
'alice' already exists, skipping
'bob' already exists, skipping
GitLab account step skipped for 'alice'.
A Personal Access Token was already issued on an earlier run (see gitlab_user_tokens.json). A re-run does not reissue one.
GitLab account and token created for 'bob'.
The token is in gitlab_user_tokens.json.
No new users added.
Files written. Each user's token is saved to gitlab_user_tokens.json
in the current directory (mode 0600) the moment it is issued:
{
"alice": "glpat-xxxxxxxxxxxxxxxxxxxx",
"bob": "glpat-yyyyyyyyyyyyyyyyyyyy"
}
Hand each user their token, and treat the file as a credential store. The registry records the account id and the issued token, which is how a re-run knows what is done; starting users get the same markers in a section of their own:
{
"users": {
"alice": {
"email": "alice@intocps.org",
"groups": [
"additional"
],
"load_balance": true,
"desired_status": "running",
"gitlab_user_id": 42,
"gitlab_pat_issued": true
}
},
"starting_users": {
"username1": {
"gitlab_user_id": 7,
"gitlab_pat_issued": true
}
}
}
Good to know:
- The issued token is named
dtaas, has theread_repositoryandwrite_repositoryscopes only, and expires after 365 days. - Password order:
--passwordor the CSVpasswordcolumn, then[[users]].passwordfor a starting user, then the hidden prompt (single user only). The CLI uses a password once, to create the account, and never saves or logs it.[[users]].passwordand the CSV column are plaintext files you keep, though: a run that reads one warns when the file is readable by other users (chmod 600it), and once the account exists it reminds you to remove[[users]].password, which stays a valid GitLab login until the user changes it.--passwordis visible in shell history andps, so prefer the prompt or the CSV. - GitLab refuses a password that is too short or built from common words
(for example one containing
passor the username). - A username or email already taken on GitLab fails the run for that user: the account is left alone and gets no token, since this run did not create it and does not know its credentials. Delete it in GitLab, or use another username or email, and run the same command again.
- Any other GitLab failure makes
user addexit non-zero, naming the users concerned; their containers stay up. - An account created on an earlier run whose token was not issued or saved needs no password on a retry: name the user again and only the token is issued.
- A token is never left live without being saved: when
gitlab_user_tokens.jsoncannot be read or written no account is created, a token that cannot be saved is revoked at once, and a retry revokes any token an earlier run issued but did not save. - When
DTAAS_GITLAB_PATand[gitlab].pathold different tokens, the exported one is used and aNote:line says so. - Run one
dtaascommand at a time per installation: the registry andgitlab_user_tokens.jsonare rewritten without a lock, so two runs at once can lose each other's changes. - For a GitLab behind an internal CA, set
[gitlab].ssl_verifyto the CA bundle's path.falseturns verification off and prints a warning. user deletenever removes the GitLab account; see user delete.
Every scenario with its full output, including each failure and how to fix it, is in the DTaaS documentation under Install with CLI, section GitLab provisioning.
For each registry user the CLI copies files/template/ into
files/<username>/, creating the directory when it is missing, and sets its
ownership to 1000:100 where it has the permission to; without that
permission it leaves the ownership as it is and carries on.
When a user has an email, the CLI automatically adds a traefik-forward-auth
routing rule for them to config/conf.server. A user added from a
users.csv row with an empty email cell gets no rule. Restart the
container for the change to take effect:
docker compose --env-file config/.env up -d --force-recreate traefik-forward-auth
Resource limits (optional)
By default each user container is created with the CPU, memory, process, and
shared-memory caps from [common.resources], merged in from the
users.resources.yml overlay. To onboard users without any caps, set
set_limits = false in that section:
# Constrained users (default): limits enforced
[common.resources]
set_limits = true
cpus = 4
mem_limit = "4G"
pids_limit = 4960
shm_size = "512m"
# Unconstrained users: no caps written, limit fields optional
[common.resources]
set_limits = false
The flag is read on every user add, so a deployment can host both constrained
and unconstrained users by toggling set_limits between runs.
Notes
user addwrites every registry user intocompose.users.ymlbut starts only the users added in that run. Users whosedesired_statusis notrunningare never started.config reconcile --fixwith an empty registry does nothing.- Usernames may include '.', '_' and '-' (must start with a letter or digit). (Whitespace, path separators, and shell metacharacters are rejected.)
- This command does not enable AuthMS authentication.
user delete
Removes one or more users from a running DTaaS instance, like userdel.
Options:
| Option | Default | Description |
|---|---|---|
USERNAMES |
— | One or more usernames to remove |
--file PATH / -f |
— | Bulk-delete users listed in a CSV (only the username column is used) |
--dry-run |
off | Preview the removal without making any changes |
Pass the usernames as arguments:
dtaas user delete username1 username2
Or bulk-delete from a CSV (the same users.csv format used by
user add --file other columns are ignored):
dtaas user delete --file users.csv
USERNAMES and --file are mutually exclusive, and one of them is required.
Each user is deprovisioned (its container stopped, its compose service and
forward-auth rule removed) and dropped from dtaas.users.registry.json. Users
that are not currently provisioned are reported and skipped, but are still
removed from the registry.
Note: the GitLab account stays.
user deleteremoves the user from the DTaaS installation and revokes the Personal Access Tokens nameddtaasthat the CLI issued to them, so they lose repository access. Their GitLab account stays, so the user can still sign in there. A revoked token is dropped fromgitlab_user_tokens.json. Revoking uses the same admin token asuser add(DTAAS_GITLAB_PATor[gitlab].pat, and[gitlab].api_url), even when[gitlab].provisionis now off. When a token cannot be revoked (GitLab unreachable, admin token missing or refused), the delete still completes, the token stays ingitlab_user_tokens.json, and the command exits with an error naming the user: revoke that token in GitLab. To remove an account, a GitLab administrator deletes it in GitLab. Do that before re-adding the same username:user addfinds the account already there, cannot know its credentials, and so issues no token for it. Adtaas.tomlstarting user is handled the same way on GitLab (token revoked, markers cleared), but its workspace is one of the installation's own, defined indocker-compose.yml, souser deletekeeps its container and forward-auth rule.
Preview a removal without making any changes with --dry-run:
dtaas user delete username1 username2 --dry-run
It lists which users would be deprovisioned and removed from the registry, then exits without stopping containers or editing any file.
The CLI automatically removes the traefik-forward-auth routing rules for
deleted users from config/conf.server. Restart the container for the change
to take effect:
docker compose --env-file config/.env up -d --force-recreate traefik-forward-auth
user pause / stop / resume
Suspend or resume specific additional (registry) users without touching
the rest of the installation, targeting one or more USERNAMES, a
--file/-f users.csv (only the username column is read, the same way
user delete does), or --all for every additional user at once.
dtaas user pause alice bob
dtaas user stop alice
dtaas user resume alice bob
dtaas user pause --file users.csv
dtaas user stop --all
To suspend or resume the whole installation, pair --all here with the
core-only dtaas platform stop/pause/resume.
| Command | docker compose verb |
Effect | Reverse with |
|---|---|---|---|
user pause |
pause |
Freeze the named users' containers (memory preserved) | user resume |
user stop |
stop |
Terminate the named users' containers, keep them | user resume |
user resume |
unpause or start, as needed |
Thaw a paused user, or restart a stopped one | — |
Each command also writes a desired_status ("paused"/"stopped"/"running")
into dtaas.users.registry.json for the users it acted on. This is what makes
the suspension durable: a later dtaas user add or dtaas config reconcile --fix checks desired_status and will not silently restart a
user you paused or stopped. Only user resume (or hand-editing the
registry) clears it back to "running".
Only additional users can be targeted here. Naming a dtaas.toml starting
user is rejected with an error: suspend or resume the whole installation
(starting users included) with dtaas platform pause/stop/resume
instead.
A username not found in the registry, or found but not currently provisioned
(e.g. user add was never run for them), is reported and skipped rather than
aborting the whole batch:
'carol' is not a registered user, skipping
alice, bob paused successfully
Options:
| Option | Default | Description |
|---|---|---|
USERNAMES |
— | One or more usernames to target |
--file PATH / -f |
— | Bulk-target users listed in a CSV (only the username column is used) |
--all |
off | Target every additional (registry) user; mutually exclusive with USERNAMES/--file |
config reconcile
Reports drift between dtaas.users.registry.json (which should be
provisioned) and the live compose.users.yml services (which are
provisioned).
dtaas config reconcile
It lists:
- missing registered but not currently provisioned;
- unexpected provisioned but not in the registry (investigate: may be a manual edit or a partial delete);
- drifted provisioned, but the live config no longer matches what
.dtaas.state.jsonrecorded when it was last provisioned; - absent provisioned (the user has a compose service), but the container
is gone, for example after an interrupted
user add; - desired-status drift provisioned, but the live container state does not
match the user's registry
desired_status(e.g.desired 'paused' but container is 'running').
When everything matches, it prints In sync: no drift detected. If Docker is
unreachable, it warns that container state could not be verified, reports
only the drift it can read from the compose file, and never reports
In sync.
Without --fix, this is read-only. Pass --fix to reprovision missing and
drifted users, restart the containers of absent users, and
pause/stop/start every provisioned user to match its desired_status:
dtaas config reconcile --fix
--fix runs the same code as dtaas user add, so it acts on the current
directory: combining it with an --output-dir other than the current
directory is rejected. It also refuses to run when Docker is unreachable.
unexpected services are never touched by --fix removing something
that's actually running is a deliberate action use
dtaas user delete for those.
Options:
| Option | Default | Description |
|---|---|---|
--output-dir PATH |
. |
Installation directory to inspect |
--fix |
off | Reprovision missing/drifted registry users after reporting |
User files
User management spans three files, each with a single owner, modelled on the
config/state split Terraform uses for .tf vs terraform.tfstate:
| File | Owner | Contents | Git |
|---|---|---|---|
dtaas.toml [[users]] |
Human, at install time | Starting users: one self-contained record per user (username, email, groups, load_balance, optional password) |
Hand-edited; keep out of Git, as it can hold credentials |
dtaas.users.registry.json |
CLI (user add / delete / pause / stop / resume) |
Additional users under users: the same fields, plus desired_status (running/paused/stopped) and any GitLab markers. Its starting_users section holds the GitLab markers alone, for dtaas.toml starting users that user add has provisioned on GitLab |
Tracked CLI-written, never hand-edited |
.dtaas.state.json |
CLI, at provisioning time | Observed runtime facts: container id, status, provisioned-at, config hash | Ignored runtime cache |
dtaas.tomlis written once by a human and never rewritten by the CLI, so a comment-bearing, reviewed config is never silently mutated.dtaas.users.registry.jsonis a database the CLI owns and mutates atomically (the wayuseraddowns/etc/passwd). Edit its users throughdtaas user add --file users.csv/delete/pause/stop/resume, not by hand.users.csvcopied bydtaas config generateis the human-editable bulk input that feedsadd/delete.desired_statusdefaults torunningfor a user who has never been paused or stopped, anduser add/config reconcile --fixskip starting any user whosedesired_statusis notrunning..dtaas.state.jsonis a disposable cache of what is actually running, refreshed on every add/delete. It is git-ignored and safe to delete.
Configuration Reference dtaas.toml
dtaas.toml is the single source of truth for all CLI commands. Generate a
blank template with dtaas config generate, fill it in, then confirm
it is valid with dtaas config validate before running any other command.
Which sections does my deployment need?
Server deployments
| Section | localhost |
insecure-server |
secure-server |
secure-server-gitlab |
|---|---|---|---|---|
[common] |
Required | Required | Required | Required |
[common.security] |
Not used | Not used | Required | Required |
[common.resources] |
Optional | Optional | Optional | Optional |
[gitlab] |
Optional | Optional | Optional | Optional |
[[users]] |
Optional | Required | Required | Required |
[frontend] |
Not used | Required | Required | Required |
[localhost] |
Required | Not used | Not used | Not used |
[insecure-server] |
Not used | Required | Not used | Not used |
[secure-server] |
Not used | Not used | Required | Not used |
[secure-server-gitlab] |
Not used | Not used | Not used | Required |
Workspace deployments
| Section | workspace-localhost |
workspace-secure-server |
|---|---|---|
[common] |
Required | Required |
[common.security] |
Not used | Required |
[common.resources] |
Optional | Optional |
[gitlab] |
Optional | Optional |
[[users]] |
Optional | Required |
[workspace-localhost] |
Required | Not used |
[workspace-secure-server] |
Not used | Required |
Annotated dtaas.toml
The full file below shows every possible key with inline comments. Copy it as a starting point and delete sections that do not apply to your deployment type (see matrix above).
# ── Metadata (top level) ──────────────────────────────────────────────────────
# Describes the instance. Only git-repo is checked by validate (an http(s) URL).
name = "Digital Twin as a Service (DTaaS)"
version = "2.2.0"
owner = "The INTO-CPS-Association"
git-repo = "https://github.com/into-cps-association/DTaaS.git"
# ── Common settings (all deployment types) ────────────────────────────────────
[common]
# Public hostname of the server. Use "localhost" for local deployments,
# a fully-qualified domain name (e.g. "dtaas.example.com") for servers,
# or a bare IP address.
server-dns = "dtaas.example.com"
# Absolute path to the DTaaS installation directory on the deployment host.
# Must exist before running validate.
path = "/opt/dtaas"
# ── TLS settings (required for: secure-server, secure-server-gitlab,
# workspace-secure-server) ──────────────────────
[common.security]
tls = true
# Absolute path to the directory containing fullchain.pem and privkey.pem.
# Used by deployment generate (seeds certs/) and platform update --certs.
certs-src = "/etc/letsencrypt/live/dtaas.example.com"
# ── Per-user container resource limits (optional, all types) ──────────────────
[common.resources]
# Enforce the limits below. Set to false to add users without any caps.
# The 4 fields are then optional and ignored. Defaults to true when omitted.
set_limits = true
cpus = 4 # CPU cores; may be fractional, e.g. 0.5
mem_limit = "4G" # memory limit unit required: G, m, k …
pids_limit = 4960 # maximum number of processes per container (integer)
shm_size = "512m" # shared memory unit required
# ── Starting users (all deployment types) ─────────────────────────────────────
# One self-contained [[users]] block per user, hand-edited once at install
# time. Presence in this file is the desired state there are no add/delete
# lists. Additional users added later with `dtaas user add` live in the
# CLI-owned dtaas.users.registry.json instead.
# Usernames must match GitLab accounts and be unique across the array.
#
# email enables traefik-forward-auth routing rules automatically;
# groups/load_balance carry per-user tags. password is optional: it is the
# initial GitLab password when `dtaas user add <username>` provisions this
# starting user's account. Avoid committing a real secret here; prefer
# supplying it at runtime instead.
[[users]]
username = "alice"
email = "alice@example.com"
groups = ["default", "dtaas"]
load_balance = true
[[users]]
username = "bob"
email = "bob@example.com"
groups = ["default", "dtaas"]
load_balance = false
# password = "Kq7v-Rm3x-Zt9w" # initial GitLab password, see above
# ── React web client OAuth app (insecure-server, secure-server,
# secure-server-gitlab) ────────────────────────
# This is a SEPARATE OAuth application from the traefik-forward-auth app
# configured in [insecure-server] / [secure-server] below.
# Redirect URI: https://<server-dns>/Library
# Scopes: api read_user read_repository openid profile Type: not Confidential
[frontend]
react-app-client-id = "dtaas-client"
react-app-oauth-url = "https://gitlab.example.com"
# ── GitLab user provisioning (optional, all types) ───────────────────────────
# Off by default. When provision = true, `dtaas user add` also creates
# each new user's GitLab account and a Personal Access Token (see
# "Add Users" in cli.md). The provisioning token must belong to an
# administrator and carry the api scope (plus admin_mode on an instance with
# Admin Mode enabled); supply it via the DTAAS_GITLAB_PAT environment variable
# rather than committing it here. A [gitlab].pat key is the fallback when the
# variable is unset or empty. ssl_verify may be a CA-bundle path for a GitLab
# behind an internal CA; false disables verification (a warning is printed)
# and true (default) uses the system trust store.
[gitlab]
provision = false
api_url = "https://gitlab.example.com"
# pat = "glpat-xxxxxxxxxxxxxxxxxxxx" # prefer DTAAS_GITLAB_PAT
# ssl_verify = "/etc/ssl/certs/corp-ca.pem" # or true (default) / false
# ── localhost deployment (dev / demo only) ────────────────────────────────────
# GitLab OAuth app for the web client; auth-authority is the GitLab URL.
[localhost]
default-user = "alice"
client-id = "dtaas-local"
auth-authority = "https://gitlab.example.com"
# ── insecure-server deployment (HTTP, demo only not internet-facing) ─────────
# GitLab OAuth app for traefik-forward-auth.
# Redirect URI: http://<server-dns>/_oauth
# Scopes: openid profile read_user Type: Confidential
[insecure-server]
oauth-url = "https://gitlab.example.com"
oauth-client-id = "abc123"
oauth-client-secret = "s3cr3t"
oauth-secret = "random-signing-string" # random; used to sign session cookies
# ── secure-server deployment (HTTPS/TLS production-ready) ───────────────────
# Same GitLab OAuth app as insecure-server, with Redirect URI using https.
[secure-server]
oauth-url = "https://gitlab.example.com"
oauth-client-id = "abc123"
oauth-client-secret = "s3cr3t"
oauth-secret = "random-signing-string"
# ── secure-server-gitlab deployment (bundled GitLab production-ready) ───────
# oauth-url points at the bundled GitLab, served under /gitlab.
[secure-server-gitlab]
oauth-url = "https://dtaas.example.com/gitlab"
oauth-client-id = "abc123"
oauth-client-secret = "s3cr3t"
oauth-secret = "random-signing-string"
# ── workspace-localhost deployment (Dex on localhost dev / demo only) ───────
[workspace-localhost]
default-user = "alice"
client-id = "workspace-local"
auth-authority = "http://localhost:5556/dex"
# ── workspace-secure-server deployment (Keycloak production-ready) ──────────
[workspace-secure-server]
keycloak-admin = "admin"
keycloak-admin-password = "change-me"
keycloak-realm = "dtaas"
keycloak-issuer-url = "https://keycloak.example.com/realms/dtaas"
keycloak-client-id = "workspace"
keycloak-client-secret = "s3cr3t"
oauth-secret = "random-signing-string"
client-id = "dtaas-frontend"
auth-authority = "https://keycloak.example.com/realms/dtaas"
Metadata
Release files for dtaas 2.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| dtaas-2.2.0.tar.gz | 165.6 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| dtaas-2.2.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 388.0 kB
Release files / dtaas-2.2.0.tar.gz
| Download URL | dtaas-2.2.0.tar.gz |
|---|---|
| Size | 165.6 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
d1937101cbed860f2a42cdf15a38d97c3bace648c85941c61a73519cf64ccc1d
|
|
BLAKE2b-256 checksum How to use checksums |
fa7fd9bf94c73387bc1b499568ad88181a4181aaecf7d3b7493b3cb8369c7840
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.
Transparency logRelease files / dtaas-2.2.0-py3-none-any.whl
| Download URL | dtaas-2.2.0-py3-none-any.whl |
|---|---|
| Size | 222.4 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
2bf8cff01efba56edf29e37308ca5d296cf1b80b30917a5a596171e5597caa63
|
|
BLAKE2b-256 checksum How to use checksums |
90fe0b520ec6447d7ce188c910e51c10f6a8ee190be1bc33fe1e4369ca5c9ce7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.
Transparency log