Docker software stack for MBON server serving early-alert dashboards.
For more detailed documentation see documentation/.
All dashboards are served from a single stack at https://mbon-dashboards.marine.usf.edu,
routed by URL path in nginx/nginx.conf:
| Path | Dashboard | Notes |
|---|---|---|
/seus, /grnms |
SE-US / Gray's Reef NMS Early Alert | data from MBON, NERRS, & GR NMS |
/fknms, /fk |
Florida Keys NMS Early Alert | satellite, buoy, & river-discharge data |
/fgbnms |
Flower Garden Banks NMS Sentinel Sites | detects anomalous satellite chlorophyll-a around the FGB reefs |
/fwc |
FWC Coral Disease (SCTLD) | project ended; route retained |
Historically each dashboard was a separate client-* git branch deployed on its
own machine. Those branches are kept for history, but the current deployment is
one unified stack on main.
docker-compose.yml handles most configuration.
The subdirectories (eg grafana/, nginx/, airflow/) contain container-specific files.
In general you will configure your stack by:
- Modifying
docker-compose.ymland container-specific configuration files inside of./grafana/,./nginx/,./airflow/, etc. Directions for this is included within aREADME.mdfile inside of each directory. Changes made to these files should be version controlled using git. - Setting up docker volumes so that data can persist when containers are recreated. Volumes are generally too large for git, so you should manage backups of these directories. A strategy for this is not included.
- Setting passwords & configuration in
.env(copy.env.exampleas a starting point). This file should not be added to git for security reasons. A backup strategy for this file is not included.
502: bad gatewayerror: this means that grafana can't connect to the timeseries database (influxDB).- check {URL}:8086/health to ensure InfluxDB okay
- check connection urls in the config, firewalls, etc
- check airflow jobs to ensure ingest is working.
- Docker Engine with the Compose v2 plugin (invoked as
docker compose; the legacy hyphenateddocker-composev1 is not supported). - ~8 GB RAM minimum for the full stack (InfluxDB + Airflow workers + Grafana + Postgres/Redis).