Configuration Reference

This page documents every significant property in the Viglet Dumont application.yaml file. Any property can be overridden via environment variables, JVM system properties (-Dkey=value), or a separate application-production.yaml file.

Override pattern

To override a property at runtime, use the environment variable convention: replace . and - with _ and uppercase everything. For example, dumont.indexing.provider becomes DUMONT_INDEXING_PROVIDER=solr.

---

Full Default Configuration

spring:
  datasource:
    url: jdbc:h2:file:./store/db/dumontDB;DATABASE_TO_UPPER=false;CASE_INSENSITIVE_IDENTIFIERS=true
    username: sa
    password: ""
    driver-class-name: org.h2.Driver
  h2:
    console:
      enabled: false
      path: /h2
      settings:
        web-allow-others: true
  jpa:
    properties:
      hibernate:
        "[globally_quoted_identifiers]": true
        "[enable_lazy_load_no_trans]": true
    hibernate:
      ddl-auto: update
    show-sql: false
  artemis:
    mode: embedded
    broker-url: localhost:61616
    embedded:
      enabled: true
      persistent: true
      data-directory: store/queue
      queues: connector-indexing.queue
  jms:
    template:
      default-destination: connector-indexing.queue

server:
  port: ${PORT:30130}

dumont:
  indexing:
    provider: turing
    job:
      batch-size: 50
    solr:
      url: http://localhost:8983/solr
      collection: dumont
    elasticsearch:
      url: http://localhost:9200
      index: dumont
      username: ~
      password: ~

turing:
  url: http://localhost:2700
  apiKey: ""

  aem.querybuilder: false
  aem.querybuilder.parallelism: 10

logging:
  file:
    name: store/logs/dum-connector.log
  level:
    com:
      viglet: INFO

---

Property Reference

Server

Property	Default	Description
server.port	30130	HTTP port for the Dumont application

Admin Account

Property	Default	Description
dumont.admin.password	(unset)	Password for the local admin account, applied at startup while no password is stored. Set it (env DUMONT_ADMIN_PASSWORD, at least 6 characters) and a container or CI start comes up with a working login and no first-access setup screen. Once a password is stored, for example set in the console, the property is ignored and a log line says so.

LLM Enrichment

Property	Default	Description
dumont.enrichment.llm.enabled	false	Lets connectors classify with an LLM through Viglet Turing (/api/v2/summary, using the configured turing.url and turing.apiKey). It also lets them extract lists from a record's text, such as the skills a job description names; only items that appear in the text are published. When off, connectors that support an LLM fallback classify with their own rules and dictionaries only.

Run Alerts

A run can finish and still carry a warning: a source reports something an operator should act on, or a field's coverage dropped sharply since the last run. The console marks such a run with a Warning badge on the Indexing Stats page.

Property	Default	Description
dumont.alerts.webhook-url	(unset)	URL that receives a JSON POST when a source's run starts carrying a warning (event, source, provider, operationType, endTime, documentCount, partial, warnings). Sent once when a source goes from healthy to warned, not on every warned run. The Turing API key is never sent. Unset, nothing is sent.

Run Metrics (Prometheus)

Each source's last run is published for Prometheus at /actuator/prometheus. The endpoint is served on a separate management port only, never on server.port, and needs no login, so keep that port on your internal network. No other actuator endpoint is exposed.

Property	Default	Description
management.server.port	30131	Port that serves /actuator/prometheus.

Every metric carries the source and provider labels:

Metric	Meaning
dumont_run_last_status	1 if the source's last run completed, 0 if it stopped short (a partial run)
dumont_run_last_finished_seconds	When the last run ended, in Unix seconds
dumont_run_last_duration_seconds	How long the last run took
dumont_run_last_documents	Documents the last run indexed
dumont_runs_total	Runs that ended, by outcome (complete, warned, partial)

The values are restored from the stored run history at startup, so a restart does not hide a source that has stopped running. Example alert rules:

- alert: DumontRunFailed
  expr: dumont_run_last_status == 0
- alert: DumontRunStalled
  expr: time() - dumont_run_last_finished_seconds > 2 * 3600

Database

Property	Default	Description
spring.datasource.url	jdbc:h2:file:./store/db/dumontDB;CASE_INSENSITIVE_IDENTIFIERS=TRUE	JDBC URL for the indexing state database. If you override it with another H2 URL, keep CASE_INSENSITIVE_IDENTIFIERS=TRUE: without it, indexing runs fail when they finish.
spring.datasource.username	sa	Database username
spring.datasource.password	""	Database password
spring.datasource.driver-class-name	org.h2.Driver	JDBC driver class
spring.h2.console.enabled	false	Enable H2 web console: do not enable in production

Message Queue (Apache Artemis)

Property	Default	Description
spring.artemis.mode	embedded	embedded (default) or native (external broker)
spring.artemis.broker-url	localhost:61616	Broker URL when using native mode
spring.artemis.embedded.enabled	true	Enable the embedded broker
spring.artemis.embedded.persistent	true	Persist queue messages to disk (survives restarts)
spring.artemis.embedded.data-directory	store/queue	Directory for persisted queue data
spring.artemis.embedded.queues	connector-indexing.queue	Queue name created on startup

Indexing Configuration

Property	Default	Description
dumont.indexing.provider	turing	Output target: turing, solr, or elasticsearch
dumont.indexing.job.batch-size	50	Number of Job Items per batch before queue delivery
dumont.indexing.provision	true	Send each source's declared field schema to the backend before its first document, so fields are created as declared rather than inferred from whichever value arrives first. A rejected schema is logged and indexing continues with the site's current schema. Turing only: other backends have no declarative schema endpoint

Extraction Resilience

An incomplete indexing run must not de-index what it never reached. Skipping that sweep is unconditional; these tune when a run is judged incomplete, and are read by the AEM connector, whose extraction is one HTTP fetch per page. See Core Concepts → Data Quality.

Property	Default	Description
dumont.reactive.indexing	false	Extract on a bounded pool instead of sequentially
dumont.reactive.parallelism	10	Threads used for parallel extraction
dumont.reactive.extractTimeoutMs	0	Per-extract wall-clock bound in milliseconds; the overrun is cancelled and counted as one failed extraction. 0 disables it: no extraction is ever cancelled for time
dumont.reactive.failureRatioThreshold	0.5	Finish standalone (skip the de-index sweep) when this fraction of a run's extractions fail. 0 disables the guard

Content Audit

The audit runs on a cron, re-enumerates every source, and files a record for anything the connector never queued. With dumont.audit.index.enabled it also compares against the search index itself and reports three differences: missing (in the source, never landed), stale (landed, but a later write failed), and orphaned (in the index, gone from the source). Missing and stale ids are requeued through the normal pipeline. Orphans are never deleted by the audit: a discovery pass that under-reports ids would otherwise take live content out of the index. Instead each pass writes the ids it found into a dated orphan report you review and apply yourself, through the orphan endpoints.

Property	Default	Description
dumont.audit.cron	0 0 3 * * *	Cron for the scheduled content audit. - disables it
dumont.audit.cron.zone	UTC	Timezone the cron is evaluated in
dumont.audit.index.enabled	true	Also compare against the search index. Requires a backend that can enumerate what it holds. Turing can; Solr and Elasticsearch report nothing and the comparison is skipped
dumont.audit.index.orphanTtl	PT24H	How long an orphan report stays applicable (ISO-8601 duration). Past it, applying is refused and a fresh audit must re-establish the report. PT0S or below switches expiry off

Turing Connection

Property	Default	Description
turing.url	http://localhost:2700	Base URL of the Turing instance
turing.apiKey	""	API Token for authenticating with Turing

Solr Direct Connection

Property	Default	Description
dumont.indexing.solr.url	http://localhost:8983/solr	Apache Solr base URL
dumont.indexing.solr.collection	dumont	Solr collection name

Elasticsearch Direct Connection

Property	Default	Description
dumont.indexing.elasticsearch.url	http://localhost:9200	Elasticsearch base URL
dumont.indexing.elasticsearch.index	dumont	Elasticsearch index name
dumont.indexing.elasticsearch.username	(none)	Optional authentication username
dumont.indexing.elasticsearch.password	(none)	Optional authentication password

AEM QueryBuilder

Property	Default	Description
dumont.aem.querybuilder	false	Enable QueryBuilder-based content discovery instead of tree traversal during full indexing
dumont.aem.querybuilder.parallelism	10	Number of parallel threads for processing discovered paths

AEM Adobe I/O Events

See AEM Connector → Adobe I/O Events.

Property	Default	Description
dumont.aem.webhook.adobe-io.client-id	–	API Key (Client ID) of the OAuth Server-to-Server credential. When set, an incoming webhook payload whose recipient_client_id does not match is rejected with 401; unset leaves the endpoint open and logs a warning at startup
dumont.aem.journal.enabled	false	Pull the same events from the Adobe I/O journal, recovering any the webhook dropped. Does not affect the webhook
dumont.aem.journal.poll-ms	60000	Delay between the end of one journal poll and the start of the next
dumont.aem.journal.batch-size	100	Events requested per journal page
dumont.aem.journal.max-pages-per-poll	20	Journal pages followed in one poll, so a large backlog drains across polls rather than in one burst
dumont.aem.journal.client-id	–	OAuth Server-to-Server API Key (Client ID) used for the journal
dumont.aem.journal.client-secret	–	OAuth Server-to-Server Client Secret used for the journal
dumont.aem.journal.ims-url	https://ims-na1.adobelogin.com/ims/token/v3	Adobe IMS token endpoint
dumont.aem.journal.scopes	openid,AdobeID,adobeio_api,event_receiver_api	Scopes requested for the token
dumont.aem.journal.sources.<source>	–	Journal URL of the event registration, per Dumont source name

AEM Fetch Strategy

Configured per source in the console, not by property: see AEM Connector → Fetch Strategy.

Dependency Tracking

Property	Default	Description
dumont.dependencies.enabled	false	Persist /content/* references extracted from each indexed page and cascade re-index any dependent document on standalone updates. Currently populated only by the AEM connector, see AEM Connector → Dependency Tracking

Logging

Property	Default	Description
logging.file.name	store/logs/dum-connector.log	Log file path
logging.level.com.viglet	INFO	Log level for Dumont application code

Indexing Log

Separate from the application log above. Every document Dumont processes leaves a row per status it passes through (prepared, sent to the queue, delivered to the search engine), each naming the document, the source that produced it and the run it belonged to. Queried through the Logging API.

It is off by default: with no engine configured, the endpoint answers with an empty page and nothing is written. Pick MongoDB when you intend to query the log, and Redis when you mostly want to tail it: the Redis engine filters and sorts in memory after reading the whole list, so it degrades as the log grows.

Turning it on is one setting. dumont.logging.engine selects both the side that reads the log and the Logback profile that writes it: logging.config is derived from it, so mongodb selects logback-spring-mongo.xml and redis selects logback-spring-redis.xml:

dumont:
  logging:
    engine: mongodb
  mongodb:
    enabled: true
    uri: mongodb://mongo:27017

Every profile writes to the console and the log file as well as its store, so choosing an engine does not cost you either. In particular there is no separate console-only profile to select in a container: the Docker image writes to stdout out of the box and the engine still applies.

Overriding the Logback profile

Set logging.config yourself only to point at a Logback file of your own; it takes precedence over the derived value and its contents are your business.

Setting it to one of the shipped profiles that writes a different store (including logback-spring-console.xml, which writes none) is refused at startup, with both settings quoted. That combination has no useful meaning: one store would be read and another, or none, written. Before 2026.3.7 it was accepted and produced a log the API reported as healthy and permanently empty. If an existing configuration has it, delete the logging.config line.

Upgrading from before 2026.3.7 in Docker

The images used to pin -Dlogging.config=classpath:logback-spring-console.xml, which switched the indexing log off no matter what else you set, so the log could not be enabled in a container at all. That pin is gone. If you added logging.config to your own compose file or entrypoint to work around it, remove it: the engine now selects the profile, and keeping the override will be refused at startup.

Property	Default	Description
dumont.logging.engine	none	none, mongodb or redis
dumont.logging.database	dumontLog	MongoDB database name
dumont.logging.collection.indexing	indexing	MongoDB collection holding the indexing rows
dumont.mongodb.enabled	false	Must also be true for the mongodb engine to activate
dumont.mongodb.uri	mongodb://localhost:27017	MongoDB connection string
dumont.redis.enabled	false	Must also be true for the redis engine to activate
dumont.redis.uri	redis://localhost:6379/0	Redis connection string
dumont.redis.key.indexing	dumontLog:indexing	Redis list key holding the indexing rows
dumont.redis.maxEntries.indexing	50000	Trim the indexing list to this length on write

The same logging.config also persists the server log (the connector's own WARN and ERROR lines) to that store, in a second collection or list. It is written by the same machinery and fails the same way, but it is not exposed through the Logging API; read it in the store, or in store/logs/dum-connector.log.

Property	Default	Description
dumont.logging.collection.server	server	MongoDB collection holding the server-log rows
dumont.redis.key.server	dumontLog:server	Redis list key holding the server-log rows
dumont.redis.maxEntries.server	10000	Trim the server list to this length on write

Each store is connected once and the connection is held for the lifetime of the connector (a MongoDB client, or a Redis pool) rather than dialled again per request. Tune either through its connection string; the usual options apply.

When the store refuses a write, the row is lost: the connector will not stall an indexing run waiting for its own log. It says so instead, at WARN in the application log (store/logs/dum-connector.log), naming both which log is losing rows and which store refused them:

Indexing log: a row could not be written to MongoDB. Rows are being lost while this lasts,
so the history will read short rather than empty.

Server log: a row could not be written to Redis. Rows are being lost while this lasts,
so the history will read short rather than empty.

An outage reports once rather than once per row, with a running count while it lasts and a final tally when writes resume, so the number of rows missing is a number you can read rather than infer. Grep for Indexing log: and Server log: after an outage before concluding that a short history means a quiet connector.

These warnings always reach the file and the console, never the store being complained about, so a MongoDB outage does not swallow the notice that MongoDB is down.

One default is Dumont's rather than the driver's. Where dumont.mongodb.uri does not set serverSelectionTimeoutMS, the indexing log applies 5000 ms in place of the driver's 30 s, so that a MongoDB which is down makes logging/status answer in five seconds rather than holding the request open for half a minute. Set the option explicitly in the URI to choose your own:

dumont.mongodb.uri: mongodb://mongo:27017/?serverSelectionTimeoutMS=2000

---

Common Production Overrides

A minimal production override:

spring:
  datasource:
    url: jdbc:postgresql://db-host:5432/dumont_db
    username: dumont
    password: strong_password
    driver-class-name: org.postgresql.Driver
  h2:
    console:
      enabled: false

dumont:
  indexing:
    provider: turing

turing:
  url: https://search.yourcompany.com
  apiKey: your-turing-api-token

server:
  port: 30130

---