Warren
Reference

Configuration reference

CommunityTeamPro

All settings are environment variables; several clusters, users and OIDC are easier in an application.yml mounted to /app/config/ (see Several clusters and users below and Single sign-on).

RabbitMQ

VariableDefaultWhat it is
WARREN_RABBIT_MANAGEMENT_URLhttp://localhost:15672the management HTTP API
WARREN_RABBIT_HOST / WARREN_RABBIT_PORTlocalhost / 5672AMQP, used for replays
WARREN_RABBIT_USERNAME / WARREN_RABBIT_PASSWORDguest / guestneeds the management tag and read and write on the vhost; monitoring for disk and memory
WARREN_RABBIT_VHOST/the vhost of this cluster entry
WARREN_RABBIT_ID / WARREN_RABBIT_NAMEdefault / –id in URLs, display name
WARREN_DLQ_NAME_PATTERNdlq, dlx, dead-letter, .dead, .failed, _error, _skippednames that make a queue a dead-letter queue

Server, database, users

VariableDefaultWhat it is
WARREN_PORT8080HTTP port
WARREN_PUBLIC_URL–base URL for links in notifications
WARREN_DATA_DIR/app/dataembedded database
WARREN_DB_URL / _USERNAME / _PASSWORDembeddeda jdbc:postgresql:// URL for PostgreSQL (Pro)
WARREN_ADMIN_USERNAME / WARREN_ADMIN_PASSWORDadmin / adminthe first local admin; change it
WARREN_LICENSE_KEY / WARREN_LICENSE_FILE–a Team or Pro licence

Metrics and alerting

VariableDefaultWhat it is
WARREN_METRICS_SAMPLE_INTERVAL30show often queues are sampled and alert rules evaluated
WARREN_METRICS_RETENTION7draw samples
WARREN_METRICS_LONG_RETENTION90dPro: hourly roll-ups
WARREN_ALERTING_ENABLED / WARREN_ALERTING_RENOTIFY_AFTERtrue / 4halert evaluation, reminders while an alert fires
WARREN_FAILURE_SCAN_INTERVAL2mhow often the dead-letter queues are read for new failures, while a New failure rule is enabled
WARREN_SMTP_HOST / _PORT / _USERNAME / _PASSWORD / _FROM– / 587mail server for e-mail channels
WARREN_OUTBOUND_CONNECT_TIMEOUT / WARREN_OUTBOUND_READ_TIMEOUT5s / 10slimits on alert webhooks, PagerDuty, Opsgenie and audit forwarding; a receiver that never answers no longer holds up alerting
WARREN_SMTP_STARTTLS / WARREN_SMTP_SSLtrue / falseSTARTTLS on 587, or TLS on 465
WARREN_PROMETHEUS_TOKEN–Team, Pro: bearer token for /actuator/prometheus

Actions and messages

VariableDefaultWhat it is
WARREN_REQUIRE_DRY_RUN / WARREN_DRY_RUN_EXPIRES_AFTERtrue / 10mbulk actions only on a dry run's token, and how long it holds
WARREN_MAX_PUBLISH_BYTES32 MiBlargest body a publish accepts; the broker's max_message_size (16 MiB by default on RabbitMQ 4) still applies
WARREN_DISCARD_BACKUP_RETENTION / WARREN_DISCARD_BACKUP_MAX_BYTES7d / 100 MiBcopies of discarded and purged messages for a restore; 0 keeps none
WARREN_PEEK_CACHE_MAX_BYTES256 MiBmemory for peeked bodies
WARREN_MAX_FULL_BODY_BYTES32 MiBthe size a body is read up to for the message view; a larger one shows its beginning and says so
WARREN_LIST_CACHE_TTL3show long a cluster's queue and binding lists are shared by all open pages, checks and rules; actions in Warren clear it, 0 turns it off
WARREN_MASKING_ENABLEDtruehide sensitive values in the message view
WARREN_MASKING_FIELDSpassword,passwd,secret,token,apikey,…names of fields and headers to hide
WARREN_MASKING_VALUESEMAIL,IBAN,CARDvalue patterns hidden wherever they appear
WARREN_MASKING_ALLOW_REVEALtrueoperators and admins may show one message's values

Audit and approval

VariableDefaultWhat it is
WARREN_AUDIT_RETENTION365dage after which finished actions and resolved alert events are deleted; 0 keeps them forever (the default before 0.15)
WARREN_AUDIT_PAYLOAD_BYTES16 KiBhow much of each message body the audit log keeps
WARREN_AUDIT_MIN_RETENTION–Pro: nothing younger is ever deleted
WARREN_AUDIT_WEBHOOK_URL–Pro: every audit event as JSON
WARREN_AUDIT_SYSLOG_HOST / _PORT / _PROTOCOL– / 514 / UDPPro: every audit event as syslog
WARREN_APPROVAL_ACTIONS–Pro: REPLAY, DISCARD, PURGE that wait for a second operator
WARREN_APPROVAL_EXPIRES_AFTER24hPro: when an undecided request expires

Several clusters and users

Lists are easiest in a YAML file mounted as /app/config/application.yml, or as indexed variables such as WARREN_CLUSTERS_0_ID, WARREN_CLUSTERS_0_MANAGEMENT_URL, WARREN_USERS_0_USERNAME:

warren:
  clusters:
    - id: prod
      name: Production
      management-url: https://rabbit.prod.internal:15672
      host: rabbit.prod.internal
      username: warren
      password: ${PROD_RABBIT_PASSWORD}
      vhost: /
    - id: prod-billing
      name: Production billing
      management-url: https://rabbit.prod.internal:15672
      host: rabbit.prod.internal
      username: warren
      password: ${PROD_RABBIT_PASSWORD}
      vhost: billing
  users:
    - { username: admin, password: ${WARREN_ADMIN_PASSWORD}, roles: [ADMIN] }
    - { username: oncall, password: ${ONCALL_PASSWORD}, roles: [OPERATOR] }
    - { username: support, password: ${SUPPORT_PASSWORD}, roles: [VIEWER] }

When warren.clusters is set, the WARREN_RABBIT_* variables are ignored; when warren.users is set, WARREN_ADMIN_* is ignored. Users from configuration are imported once, when the user table is still empty. From then on manage them under Users in the UI; later changes to the configured users have no effect. If you lock yourself out, delete the rows in app_user and restart: the seed runs again.