# Pneumatic Workflow Configuration

# This is a configuration file listing all parameters and their default values.
# After changing the parameters, you need to restart the Docker containers.


# =============================================================================
# Server address
# =============================================================================
# SERVER_ADDRESS=localhost
# BACKEND_URL=http://localhost:8001
# FRONTEND_URL=http://localhost
# FORMS_URL=http://localhost/forms
# WSS_URL=ws://localhost:8001

# Custom form domain works only with SSL=yes
# If FORM_DOMAIN is specified, use NGINX_CONF_TEMPLATE=./nginx/ssl_forms_templates/
# FORM_DOMAIN=

# =============================================================================
# Nginx / SSL
# =============================================================================
# SSL=no
# NGINX_CONF_TEMPLATE=./nginx/ssl_templates/

# --- Certbot ---
# CERTBOT_ENABLE=no
# CERTBOT_EMAIL=

# =============================================================================
# Database
# =============================================================================

# --- PostgreSQL ---
# POSTGRES_PASSWORD=postgres_password
# POSTGRES_REPLICA_PASSWORD=postgres_password
# POSTGRES_HOST=postgres
# POSTGRES_USER=postgres_user
# POSTGRES_DB=postgres_db
# POSTGRES_REPLICA_HOST=postgres
# POSTGRES_REPLICA_USER=postgres_user
# POSTGRES_REPLICA_DB=postgres_db

# --- Redis ---
# Cache (db0), auth tokens (db1), websocket channels (db2), sessions (db3) and
# the event buffer (db4, see Logging / Events below).
# REDIS_PASSWORD=redis_password

# --- RabbitMQ ---
# RABBITMQ_USER=rabbitmq_user
# RABBITMQ_PASSWORD=rabbitmq_password

# =============================================================================
# Application
# =============================================================================

# ENVIRONMENT=Production
# LANGUAGE_CODE=en

# --- Backend ---
# WORKERS_COUNT=2
# DJANGO_DEBUG=no
# ADMIN_PATH=admin
# DJANGO_SECRET_KEY=django_secret_django_secret_django_secret
# ENABLE_LOGGING=no
# SENTRY_DSN=

# --- Frontend ---
# NODE_OPTIONS=--max-old-space-size=3072
# SENTRY_FRONTEND_DSN=
# MCS_RUN_ENV=prod

# --- File service ---
# SENTRY_FILE_DSN=

# =============================================================================
# SIGNUP
# =============================================================================
# SIGNUP=yes

# =============================================================================
# Authorization
# =============================================================================

# --- Microsoft Entra ID (AD) ---
# MS_AUTH=no
# MS_CLIENT_ID=
# MS_CLIENT_SECRET=
# MS_AUTHORITY=

# --- Google ---
# GOOGLE_AUTH=no
# GOOGLE_OAUTH2_CLIENT_ID=
# GOOGLE_OAUTH2_CLIENT_SECRET=
# GOOGLE_OAUTH2_REDIRECT_URI=

# --- SSO ---
# SSO_AUTH=no

#   Okta
# SSO_PROVIDER=okta
# OKTA_DOMAIN=
# OKTA_CLIENT_ID=
# OKTA_CLIENT_SECRET=
# OKTA_REDIRECT_URI=

#   Auth0
# SSO_PROVIDER=auth0
# AUTH0_CLIENT_ID=
# AUTH0_CLIENT_SECRET=
# AUTH0_DOMAIN=
# AUTH0_REDIRECT_URI=

# =============================================================================
# Email
# =============================================================================
# EMAIL=no

# --- Server ---
# EMAIL_PROVIDER=smtp
# EMAIL_HOST=
# EMAIL_PORT=
# EMAIL_HOST_USER=
# EMAIL_HOST_PASSWORD=
# EMAIL_USE_TLS=
# EMAIL_USE_SSL=
# EMAIL_TIMEOUT=

# --- Customer.io ---
# EMAIL_PROVIDER=customerio
# CIO_WEBHOOK_API_VERSION=
# CIO_WEBHOOK_API_KEY=
# CIO_TRANSACTIONAL_API_KEY=
# CIO_TEMPLATE__RESET_PASSWORD=
# CIO_TEMPLATE__INVITE=
# CIO_TEMPLATE__USER_DEACTIVATED=
# CIO_TEMPLATE__NEW_TASK=
# CIO_TEMPLATE__TASK_RETURNED=
# CIO_TEMPLATE__ACCOUNT_VERIFICATION=
# CIO_TEMPLATE__WORKFLOWS_DIGEST=
# CIO_TEMPLATE__TASKS_DIGEST=
# CIO_TEMPLATE__USER_TRANSFER=
# CIO_TEMPLATE__UNREAD_NOTIFICATIONS=
# CIO_TEMPLATE__GUEST_NEW_TASK=
# CIO_TEMPLATE__OVERDUE_TASK=
# CIO_TEMPLATE__MENTION=
# CIO_TEMPLATE__TASK_REMINDER=
# CIO_TEMPLATE__COMPLETE_WORKFLOW=
# CIO_TEMPLATE__VACATION_DELEGATION=

# =============================================================================
# Captcha
# =============================================================================
# CAPTCHA=no
# RECAPTCHA_SITE_KEY=
# RECAPTCHA_SECRET_KEY=
# RECAPTCHA_TESTING=no

# =============================================================================
# AI
# =============================================================================
# AI=no

# --- OpenAI ---
# AI_PROVIDER=openai
# OPENAI_API_KEY=
# OPENAI_API_ORG=
# AI_HTTP_PROXY=http://user:pass@proxy.example:8080
# AI_HTTPS_PROXY=http://user:pass@proxy.example:8080

# =============================================================================
# Push notifications
# =============================================================================
# PUSH=no

# --- Firebase ---
# PUSH_PROVIDER=firebase
# FIREBASE_VAPID_KEY=
# FIREBASE_API_KEY=
# FIREBASE_AUTH_DOMAIN=
# FIREBASE_PROJECT_ID=
# FIREBASE_STORAGE_BUCKET=
# FIREBASE_SENDER_ID=
# FIREBASE_APP_ID=
# FIREBASE_MEASUREMENT_ID=
# FIREBASE_PUSH_APPLICATION_CREDENTIALS=/pneumatic_backend/firebase-push.json

# =============================================================================
# Analytics
# =============================================================================
# ANALYTICS=no

# --- Segment ---
# ANALYTICS_WRITE_KEY=
# ANALYTICS_DEBUG=no

# =============================================================================
# Logging / Events
# =============================================================================
# User action events are buffered in Redis and forwarded by the OpenTelemetry
# collector to the store LOGS_BACKEND names. The journal is off by default:
# none of its containers start and nothing is written to Redis.

# otlp / elasticsearch = a store outside this machine; unset or empty =
# pipeline off (the default). Nothing that stores or shows the journal runs
# here: the collector forwards the events, the store lives on a machine of its
# own with a compose file of its own - logging/grafana for Loki + Grafana,
# logging/elasticsearch for Elasticsearch + Kibana. start.sh reads this value
# and starts the collector on its own; running compose by hand needs the
# COMPOSE_PROFILES line at the bottom of this section as well. The defaults
# of the other LOGS_* values below are in x-logs-environment of the compose
# files.
# LOGS_BACKEND=otlp
# LOGS_OTLP_ENDPOINT=http://otel-collector:4318

# The file service writes into the same journal (uploads, downloads and
# refused downloads) and reads LOGS_BACKEND, REDIS_PASSWORD and
# LOGS_STREAM_MAXLEN from this file too; there is nothing to configure for it
# separately. Deploy order on an upgrade: backend first, then the file service.
# The backend declares the file event types; a file service ahead of it would
# have its records filed under the other category with the file name exposed.

# --- Event buffer: db4 of the Redis above ---
# Events wait in a Redis stream between emit() and the collector, in db4 of the
# shared Redis; REDIS_PASSWORD opens it. The stream is the only copy of an
# event until the collector has it, and that Redis keeps RDB snapshots only,
# so a crash costs at most ~20 seconds of events.

# How many events the buffer keeps before the oldest are dropped. It is also
# the memory budget: an event costs ~700 B in the stream (plan with 1 KB), so
# the default 250 000 is ~250 MB, plus 10 MB of dead letter.
# What it buys is how long the collector may be down without losing events:
# ~69 hours at 1 event/s, ~7 hours at 10/s, ~42 minutes at 100/s. The rate of
# workflow events, the bulk of the journal, over the last day:
#   select count(*) / 86400.0 from processes_workflowevent
#   where created > now() - interval '1 day';
# LOGS_STREAM_MAXLEN=250000
# Events in one POST to the collector. A batch refused with 413 (body too
# large for the collector or a proxy in front of it) goes to the dead
# letter: lower this value.
# LOGS_CONSUMER_BATCH_SIZE=1000

# --- OTLP receiver outside this machine (LOGS_BACKEND=otlp) ---
# The Loki of logging/grafana/docker-compose.yml takes the records at
# http://<that machine>:3100/otlp; a stack of your own takes them wherever
# its OTLP/HTTP receiver listens.
# LOGS_OTLP_EXPORT_ENDPOINT=
# The whole Authorization header value ("Bearer ...", "Basic ..."); it is
# sent even when empty, see otel-collector/backend-otlp.yaml.
# LOGS_OTLP_EXPORT_AUTH=

# --- Elasticsearch (LOGS_BACKEND=elasticsearch) ---
# The store runs on a machine of its own, see logging/elasticsearch. This
# machine needs the variables below, the CA file of the store and, once, a
# queue volume the collector can write into. The collector image runs as
# uid 10001 and a fresh named volume belongs to root, so before the first
# start:
#   docker volume create pneumatic_otel-queue
#   docker run --rm -v pneumatic_otel-queue:/q alpine chown 10001:10001 /q
# Afterwards the collector log has no "permission denied", and
#   docker run --rm -v pneumatic_otel-queue:/q alpine ls -l /q/queue
# shows the queue database.

# Address of the store. https, always: this is the audit journal of every
# account in the installation, and the host name has to be one of the names
# the certificate was issued for (ES_CERT_DNS on the storage machine).
# LOGS_ELASTICSEARCH_URL=https://es.pneumatic.internal:9200

# How the collector authenticates. An API key is the way: it carries a role
# that can append to the journal and nothing else, it expires on a date of its
# own, and rotating it is this one line plus a restart - see
# logging/elasticsearch/scripts/rotate-api-key.sh, which prints the value.
# LOGS_ELASTICSEARCH_API_KEY=

# The login and password of an Elasticsearch account, as an alternative to the
# key above. Set one or the other, never both. A password is worse: it belongs
# to a person, it has no expiry date, and changing it takes the account away
# from everything else that used it.
# LOGS_ELASTICSEARCH_USER=
# LOGS_ELASTICSEARCH_PASSWORD=

# The certificate authority of the store, inside the collector container. The
# file is logging/elasticsearch/certs/ca/ca.crt on the storage machine; copy
# it to otel-collector/es-ca.crt here - that directory is already mounted
# read only, so no compose change is needed. An empty value means the system
# certificate store, for a store with a publicly trusted certificate.
# LOGS_ELASTICSEARCH_CA_FILE=/etc/otel/es-ca.crt

# The data stream the events are written to. The name is not free-form:
# `logs-<dataset>.otel-<namespace>` is what Elasticsearch recognises as an
# OTel log stream, and it is what the index template, the retention policy and
# the writer role are bound to. Change it only together with ES_DATA_STREAM
# and ES_INDEX_PATTERN on the storage machine.
# LOGS_ELASTICSEARCH_INDEX=logs-pneumatic.events.otel-default

# Where the collector keeps the queue of batches it has not delivered yet.
# The default is the mount point of the otel-queue volume of the collector;
# there is no reason to change it.
# LOGS_ELASTICSEARCH_QUEUE_DIR=/var/lib/otelcol/queue

# One thing to watch in this mode. A store that cannot be reached at all
# (connection refused, name does not resolve) makes the exporter drop the
# batch instead of retrying it, and by then the consumer has been answered
# and the records are out of the buffer. So: stop the collector before
# restarting the store, and watch the counters it publishes on :8888 -
#   docker exec pneumatic-otel-collector wget -qO- localhost:8888/metrics \
#     | grep -E 'send_failed_log_records|queue_size'
# The first counter is events lost for good, the second is how full the
# disk queue is. otel-collector/backend-elasticsearch.yaml has the details.

# --- Start the services of the mode ---
# Only for running `docker compose` by hand: start.sh derives the same from
# LOGS_BACKEND above and passes it itself. Uncomment the lines matching
# LOGS_BACKEND. Without them the collector does not start, and a pipeline that
# is on piles events up in the buffer until LOGS_STREAM_MAXLEN drops the oldest.
# otlp:
# COMPOSE_PROFILES=logs-otlp
# elasticsearch:
# COMPOSE_PROFILES=logs-elasticsearch

# =============================================================================
# File Service
# =============================================================================
# FILE_SERVICE_URL=http://localhost/files
# FILE_DEBUG=false
# FILE_CONFIG=Production
# FILE_WORKERS=1
# FILE_PORT=8000
# FILE_FASTAPI_BASE_URL=http://localhost/files
# FILE_MAX_FILE_SIZE=104857600
# FILE_BUCKET_PREFIX=pneumatic-dev-test
# FILE_CHUNK_SIZE=1048576

# --- File Service Auth (internal, must match Django backend) ---
# FILE_KEY_PREFIX_REDIS=:1:
# FILE_AUTH_TOKEN_ITERATIONS=1

# --- File Service Database ---
# FILE_POSTGRES_USER=pneumatic
# FILE_POSTGRES_PASSWORD=pneumatic
# FILE_POSTGRES_DB=pneumatic
# FILE_POSTGRES_HOST=postgres
# FILE_POSTGRES_PORT=5432

# --- File Service Storage: Local (SeaweedFS) ---
# FILE_STORAGE_TYPE=local
# FILE_SEAWEEDFS_S3_ENDPOINT=http://seaweedfs-filer:8333
# FILE_SEAWEEDFS_S3_ACCESS_KEY=any-secret-key-will-work
# FILE_SEAWEEDFS_S3_SECRET_KEY=any-secret-key-will-work
# FILE_SEAWEEDFS_S3_REGION=us-central1
# FILE_SEAWEEDFS_S3_USE_SSL=false

# --- File Service Storage: Google Cloud ---
# To use GCS, set FILE_STORAGE_TYPE=google and fill in the values below.
# FILE_GCS_S3_ENDPOINT=
# FILE_GCS_S3_ACCESS_KEY=
# FILE_GCS_S3_SECRET_KEY=
# FILE_GCS_S3_REGION=


# =============================================================================
# Billing
# =============================================================================
# BILLING=no

# --- Stripe ---
# STRIPE_SECRET_KEY=
# STRIPE_WEBHOOK_SECRET=
# STRIPE_WEBHOOK_IP_WHITELIST=