This page explains how shared storage works. In simple terms: three core nodes pool their disks so services can store data centrally, and dashboards watch storage health so issues are caught early.
Technical topology and pool details are documented below.
/mnt/cephfs/docker/...| Pattern | Storage path |
|---|---|
| VM disks | RBD-backed Proxmox storage. |
| Shared app config | CephFS paths under /mnt/cephfs/docker/<stack>/<service>/. |
| Transcode/shared scratch where needed | CephFS paths used by services such as Plex workers. |
| Bulk media and long-term retained files | Unraid NFS volumes rather than CephFS. |
The full hardware/storage split is summarized in Hardware, and Swarm storage consumption is documented in Docker Swarm.
| Symptom | Likely area to check |
|---|---|
| Many apps fail at once | CephFS mount, Swarm node storage path, or shared database service. |
| One app loses config | Its /mnt/cephfs/docker/<stack>/<service> path or permissions. |
| Writes become slow | Ceph health, network/storage traffic, or pool fill ratio. |
| A moved service will not start | Node mount state, placement constraints, or missing external volume. |
Grafana dashboards are used to watch CephFS/storage behavior alongside host CPU/RAM/temperature data.