Run IPFS on a VPS: a practical Kubo operations checklist
Plan a node around administrative boundaries, storage, monitoring, retained releases, and recovery—not just a successful daemon start.
IPFS VPS / SELF-OPERATED INFRASTRUCTURE
Your node needs an operating plan.
Running Kubo on a VPS gives you control over a node. It also gives you responsibility for its interfaces, retained releases, storage growth, and recovery.
Start with a narrow role
Decide whether the machine retains your site, serves a public gateway, or supports a private publishing workflow. List the interfaces that must be reachable and the reason for each. Do not accidentally turn a personal content provider into an unrestricted public retrieval service. Document the node’s relationship to the public domain and any independently retained copies.
Keep administration private
Kubo’s RPC API provides administrative control over the node. It is not the same as a public gateway. Keep it private by default and review the complete access path, including firewalls, reverse proxies, and container port mappings. If remote automation is needed, design a protected, narrowly scoped route rather than exposing a management port to make a tool work.
Plan the repository lifecycle
Track the current release, rollback releases, and archival copies separately. Measure disk growth and keep enough headroom for the workload you actually observe. Record which root CIDs should remain pinned and why. Maintenance should be based on that manifest, not on deleting unfamiliar files until the machine has free space. Protect sensitive node configuration separately from public artifacts.
Practice a replacement
A backup becomes useful when you have restored it successfully. Recover a known release on a replacement environment and test the public delivery path. Keep source files, build instructions, release records, and necessary private configuration in appropriate independent storage. Monitor meaningful content retrieval as well as process liveness, and give alerts a current owner who knows the recovery procedure.
Technical reference: Kubo RPC administrative-interface reference. Review the relevant configuration before applying it to a production environment.
COMMON QUESTIONS
IPFS VPS: the details
How much RAM or storage do I need?
There is no universal size for every site and gateway workload. Estimate your retained data and traffic, then measure actual resource use and leave operating headroom.
Does one VPS make the site redundant?
No. A single machine still has dependencies on its host, network, disk, and operator. Review independent retention and recovery according to the consequences of an outage.
FIELD GUIDES
Go deeper in the Website Lab
IPFS hosting, pinning, and costs: plan the whole delivery path
Compare local nodes, managed pinning, and gateway delivery without treating an upload or a cached page as an availability guarantee.