Storage Across Nodes

When you run more than one node, there is no shared filesystem between them. Each node has its own disk, and a managed volume is a directory on exactly one of them. A volume cannot be split, so ZenSailor elects one node to hold a service's data the first time it deploys successfully, and remembers that choice.

It always goes back to its data

A service with volumes redeploys onto the node holding its data, every time. It can never start on a different node against an empty directory. If that node is unavailable, the deploy waits and tells you which node it needs.

Moving data is deliberate

To move a service to another node, run a volume migration from its storage panel. The service is stopped, the data is copied through your backup repository, the copy is verified, and only then does the service move. The original copy is never deleted.

Draining will not move it

Draining the node that holds a service's data does not relocate that service — that would start it against an empty directory. It stays down until the node returns or you migrate the volume. Stateless services still reschedule freely.

Plan around the trade-off: a service with managed volumes lives on one machine, so losing that machine means restoring from backup, not failing over. If a service must move freely between nodes, keep its state outside the platform — an external database or object store — and it stays portable.