Configuration

The config seed, where the configuration is kept, git and ytt

Example
Where config.yaml is read from
Variables

Change these to decide where the configuration comes from: a config seed, the database, git or ytt templates.

Example

Apply a mounted config seed, check it every five minutes, and keep config.yaml in the database. In docker-compose.yml:

    environment:
      - CONFIG_SEED_PATH=/seed/seed.yaml
      - CONFIG_SEED_RELOAD_SECONDS=300
      - ENABLE_CONFIG_IN_DATABASE=1
      - ALLOW_ENV_EDIT=0              # the environment is declared here, not on the settings pages
    volumes:
      - ./seed:/seed:ro

An invalid seed stops the server from starting. See Config seed for the file format.

Where config.yaml is read from

The categories, roles and constants come from config.yaml on disk or from its copy in the database, decided in this order:

ENABLE_CONFIG_IN_DATABASE, when it is set, in the environment or on the Configuration tab of the settings page: 1 reads the database, 0 reads the file.
Otherwise, the configuration source stored in the database (file or database), which instances set up before 6.3.0 chose on the settings page.
Otherwise, the file: the config.yaml of the repository with Use for config enabled, else of a repository with Use for forms enabled, else the one at CONFIG_PATH.

An explicitly set ENABLE_CONFIG_IN_DATABASE always wins, even 0 over a stored database. It is read each time the configuration loads, so a change applies without a restart. The Status page shows the source in use on its Information tab.

Variables

Every variable of this group, with its type, default and description:

Variable	Choices/Defaults	Comments
CONFIG_SEED_PATH string added in version 6.3.0	Default:	Config seed file a valid file path Path to a declarative config seed. When set, the yaml file it points to is applied at startup and the admin objects it declares become read-only in the interface, so a whole instance can be rebuilt from git on an empty database. Secrets are referenced as ${ENV_VAR} and resolved from the environment, so the file itself holds none. Unrelated to CONFIG_PATH, which holds the forms configuration. An invalid seed makes the server refuse to start, and so does a path whose file is missing (a mounted file may only be late). Leaving it empty turns the seed off: the next start releases every object an earlier seed made read-only, keeping its values, so it can be changed in the interface again. See the Config seed page in the documentation.
CONFIG_SEED_RELOAD_SECONDS integer added in version 6.3.0	Default: 60	Config seed reload interval (seconds) 0 or a number of seconds How often the config seed file is checked and re-applied when its content has changed, in seconds. This is what lets a seed edited in git take effect without restarting: on kubernetes a ConfigMap is remounted under the running pod within about a minute and the change is picked up from there. A reload that fails is never fatal, unlike the one at startup: the configuration already in force is kept and the Status page reports it. Set it to 0 to check only at startup, leaving POST /api/v2/config-seed/apply and SIGHUP. Takes effect at once, without a restart. Ignored when CONFIG_SEED_PATH is empty.
ALLOW_ENV_EDIT boolean added in version 6.3.0	Default: 1	Allow editing these settings 0, 1 Whether the settings pages may write persistent/.env. Set it to 0 in a declarative deployment (kubernetes, docker-compose, ArgoCD): there the file is either ephemeral, so a save is silently lost at the next restart, or durable, so it drifts away from the manifest that is meant to be authoritative. With 0 every variable is shown with its value in force but cannot be changed, and the save endpoint refuses.
ENABLE_CONFIG_IN_DATABASE number added in version 6.1.0	Choices: 0 (default) 1	Store the configuration in the database 0, 1 If you use git repositories to store the forms, you can choose to store the master config.yaml in git as well. Or, you can choose to have the config.yaml in the database instead. Having a clean separation: master config.yaml in database, and the actual forms in git. This gives the config.yaml in the database the highest priority in the loading order.
GIT_CLONE_COMMAND string added in version 6.0.2	Default: git clone	Git clone command a valid git clone command The command used to clone git repositories. Note that branch and repository will be auto added.
GIT_PULL_COMMAND string	Default: git pull	Git pull command a valid git pull command The command used to pull git repositories. Note that branch and repository will be auto added.
GIT_PUSH_COMMAND string	Default: git push	Git push command a valid git push command The command used to push git repositories (used by the designer’s sync to git feature). Note that `-u origin HEAD` will be auto added.
USE_YTT number added in version 5.0.2	Choices: 0 (default) 1	Enable the ytt interpreter 0, 1 https://github.com/carvel-dev/ytt is a tool for yaml templating. It can be activated by this variable.
YTT_ALLOW_SYMLINK_DESTINATIONS string added in version 5.0.8	Default:	ytt allowed symlink destinations valid paths Ytt disables templating with symlinks by default. If you want to allow symlink use on specific directories, this variable can be set to the path(s) where the symlinks are allowed (multiple paths separated by Node’s `path.delimiter`). Note: enabling this may come with some risks, see ytt FAQ for more info.
YTT_DANGEROUS_ALLOW_ALL_SYMLINK_DESTINATIONS number added in version 5.0.8	Choices: 0 (default) 1	ytt dangerous allow-all symlink destinations 0, 1 Ytt disables templating with symlinks by default. If you want to allow all symlink use EVERYWHERE, you can set this variable to 1. Note: enabling this comes with some risks (hence the ‘DANGEROUS’ part), see ytt FAQ for more info.
YTT_VARS_PREFIX string added in version 5.0.8	Default:	ytt vars prefix a valid prefix Environment variable prefix for ytt to get data values from. For example, when set to `MY_PREFIX`, `MY_PREFIX_my_var=value` results to the ytt data value `my_var=value`.
YTT_LIB_DATA_{dynamic} string added in version 5.0.8	Default:	ytt lib data a valid lib data value This is not a single variable but rather an infinite amount of variables. You can set as many as you want. The dynamic part is the name of the lib data. For example `YTT_LIB_DATA_MYLIB=values.yml` results in the contents of `values.yml` to be used by ytt in the `mylib` library.