Repository navigation
Operations
Telling whether a JobWarden server is healthy, reading its logs, and what to do when something's wrong. Backups and upgrades are on Updating; every setting on Configuration.
| Address | Answers |
|---|---|
/actuator/health |
{"status":"UP"}, or DOWN with status 503 |
/actuator/health/liveness |
Whether the server runs at all: Docker's HEALTHCHECK asks it every 30 seconds |
/actuator/health/readiness |
Whether it can answer requests: the database too |
/actuator/info |
The version, the commit it was built from, the Java it runs on |
/actuator/metrics |
The metrics' names; /actuator/metrics/<name> one of them |
Metrics worth a look: http.server.requests (each page's count and time, by status), tasks.scheduled.execution (each
scheduled check's runs, time and outcome: ?tag=outcome:ERROR), hikaricp.connections.active, jvm.memory.used.
Stopping the server (Ctrl+C, docker stop) lets the requests under way finish first, for up to 20 seconds. The jar
carries its software bill of materials (CycloneDX) at META-INF/sbom/application.cdx.json, for vulnerability scanners.
One line per event, on the console:
2026-11-14T18:53:02.237+02:00 INFO 6790 --- [jobwarden] [omcat-handler-0] [3f9c2a71d04e8b65] c.j.jobs.service.impl.DefaultApplicationService : Created application 42 for profile 1 as SAVED
-
The id in brackets ties the lines of one request together; the
X-Request-Idresponse header gives it, so a failure someone reports ("it said that didn't work at 18:53") leads to its lines. Each run of a scheduled check (the inboxes, the watchlist, the reminders' cleanup) has an id of its own too. A request that comes with anX-Request-Id(from a proxy in front, say) keeps it. -
Levels: INFO for what changes, WARN for what users did that the server refused (a stale link, a broken rule),
ERROR only for what shouldn't happen, always with an error code (
[JOBS-001]) to search for. DEBUG adds the reads. - No personal data: lines name profiles and jobs by id, never by name, title or company.
-
JSON (
LOGGING_STRUCTURED_FORMAT_CONSOLE=ecs): the id is therequestIdfield, the codeerrorCode.
| What | Why, and what to do |
|---|---|
Detected resolved migration not applied to database at start |
A migration was added while the server ran from sources, before an earlier one existed. Apply it with Flyway's outOfOrder, or restore the backup |
The database refused the password changed in Settings in the log at start |
PostgreSQL doesn't take the password kept in ~/.jobwarden/database.password (an older backup restored, say), so the server connects with the configured one. Change it again in Settings, or delete the file |
password authentication failed at start |
Neither the password changed in Settings nor the configured one is PostgreSQL's: set SPRING_DATASOURCE_PASSWORD to the one it has, or reset it where PostgreSQL runs: in Docker, docker compose exec postgres psql -U jobwarden -c "ALTER ROLE jobwarden PASSWORD '...'", which the container lets in without one |
jobwarden.secret-key must be 32 bytes in Base64 at start |
JOBWARDEN_SECRET_KEY is set, but isn't a key: openssl rand -base64 32 makes one (and the secrets kept with the old one need entering again) |
| A mailbox says it couldn't be read | The inbox page says why; the log has the line, with the profile's id. An app password, not the account's, is what most providers need |
| Run is refused on the AI Assistant page | The card says why: the assistant isn't installed or signed in on this computer, or too many runs are going on. claude auth status / codex login status tell |
| A page says "That didn't work" | The response had an error: find its X-Request-Id in the browser's network tab, then its lines in the log |
Still stuck? Ask on Discord, or open an issue with the version, what you did and the log's lines.
JobWarden · Releases · Discord · Suggest a feature · Report a bug · MIT License