wireops demo

Sample self-hosted stacks and a scheduled job used to populate a wireops demo environment for screenshots and docs. Not meant for production use — image tags and ports are chosen for a local screenshot pass, not for real deploys.

Stacks

• stacks/uptime-monitor — Uptime Kuma, single service, minimal wireops.yaml.
• stacks/docs-site — nginx + whoami, static page injected via inline configs: (no bind mounts — the worker only ever receives the rendered compose file, never the rest of the repo, so relative bind mounts to sibling repo files don't resolve on deploy).
• stacks/metrics-stack — Prometheus + Grafana + node-exporter, multi-service with named volumes, an inline configs: scrape config, and a worker.tags example.
• stacks/gitea — Gitea + Postgres + Redis + a runner, 4 services on 2 networks with a depends_on chain (runner → gitea → db/cache) and 4 named volumes — good for showing off the dependency graph.
• stacks/paperless — Paperless-ngx + Postgres + Redis + Gotenberg + Tika, 5 services, 2 networks, 6 named volumes, deepest dependency chain of the set.
• stacks/nginx-configs-demo — nginx serving a static page. No wireops.yaml involvement at all: the service annotates dev.wireops.config.<name>: <repo-relative source>:<in-container target> directly, and wireops resolves the source straight from this repo checkout and synthesizes the compose configs: block into the rendered file at deploy time. Two annotations show both source shapes: nginx-conf maps a single file (files/nginx.conf), and html maps a whole directory (files/html/) — a directory source is expanded recursively, every file under it becomes its own config entry mounted under the declared target. Unlike docs-site/metrics-stack above (which hand-author a native configs: block with inline content directly in docker-compose.yml), this raw compose file never mentions configs: at all — no bind mounts, no stub/placeholder entries, no copy-pasting config content into YAML.

Jobs

• jobs/nightly-backup — daily, once, tars a volume and prunes old backups.
• jobs/healthcheck-ping — every 15 min, once_all (fans out to every matching worker), curls each demo stack's exposed port.
• jobs/gitea-db-vacuum — weekly, once, runs VACUUM ANALYZE against the gitea stack's Postgres, targets the gitea stack's compose network directly.
• jobs/prune-docker — weekly, once_all, demo docker cleanup job.
• jobs/config-demo-report — daily, once, demonstrates job.yaml's configs: (name + path + target): files/report.sh is a plain git file, resolved server-side and bind-mounted read-only into the job container at /scripts/report.sh — the worker never clones this repo, it only receives the resolved script content in the dispatch command.

All stacks and jobs target worker.tags: [local, node] — match those tags on whichever worker you connect to this demo.

Point a wireops repository record at this repo, add a stack per stacks/* subdirectory (or a wildcard import), and schedule each jobs/*/job.yaml.