What is ADAMANT IPFS Node?

ADAMANT IPFS Node is a self-hosted content-addressed file storage service for application file delivery. It runs as a standalone Node.js service embedding Helia and libp2p with an Express REST API. It avoids the public DHT and Kubo API. Instead, it forms a controlled peer mesh over TCP using Noise and Yamux, where nodes exchange blocks and health data only with configured peers. Features include bounded disk usage with watermark garbage collection, deterministic replica placement via rendezvous hashing with age-based tiering, cross-node repair cycles, health checkpoints with monotonic height, and rate limits with admission control. Configured via a single JSON5 file. Deployable via Docker, Docker Compose, or source on Node.js 24. Requires a TLS reverse proxy and admin API key for production. Used by ADAMANT Messenger for attachment delivery.

Should you self-host ADAMANT IPFS Node?

ADAMANT IPFS Node scores 48/100 on selfhost.directory (how scoring works) and ranks #8 of 15 in File Storage. It has 307 GitHub stars.

It shipped 1 stable release in the past year. The latest, v0.1.0, came out on 30 September 2026.

Setup is advanced: plan for several services and some ongoing maintenance. An official Docker image is available. It is GPL-3.0-licensed and written mostly in Typescript.

Its closest alternative on selfhost.directory is juicefs (73/100, 14 582 stars). It scores higher overall, so compare the two before you pick. See ADAMANT IPFS Node vs juicefs side by side.

Features

Administrative API key authentication
Bounded disk usage
Controlled peer mesh
CORS Configuration
Cross-node repair
Deterministic replication
Graceful Shutdown
Health checkpoints
Health monitoring endpoint
Rate limits and admission control
REST API for file upload/download
Storage metrics and policy API
Structured JSON logging
Trusted proxy support

How to configure ADAMANT IPFS Node

adminApiKey''Secret required in the x-api-key header on administrative routes
Secret required in the x-api-key header on administrative routes
values: string (empty or >= 32 chars)
cors.allowedOriginsBrowser origins accepted by the API
Browser origins accepted by the API
values: array of strings (exact origins, subdomain wildcards, or app:// desktop origins)
diskUsageScanPeriodHow often the storage scan that feeds the metrics runs
How often the storage scan that feeds the metrics runs
values: cron string
downloadIdleTimeoutfindFileTimeoutMaximum pause between download chunks, in milliseconds
Maximum pause between download chunks, in milliseconds
values: integer >= 1
downloadMaxDurationMsmax(14400000, size-derived)Absolute ceiling for one download response, in milliseconds
Absolute ceiling for one download response, in milliseconds
values: integer >= 1
downloadMinBytesPerSecond32768Rate used to derive a size-aware complete-transfer deadline
Rate used to derive a size-aware complete-transfer deadline
values: integer >= 1
enableDebugApifalseMounts /api/debug/*, which still requires the key
Mounts /api/debug/*, which still requires the key
values: true|false
findFileTimeoutMilliseconds allowed for locating a file before a download fails
Milliseconds allowed for locating a file before a download fails
values: integer >= 1
health.checkpointIntervalMs60000Length of one checkpoint round
Length of one checkpoint round
values: integer >= 1000
health.clockSkewToleranceMs10000Allowed clock difference between two attesting peers
Allowed clock difference between two attesting peers
values: integer >= 0
health.maxCheckpointAgeMscheckpointIntervalMs * 3Age at which the last completed checkpoint becomes stale
Age at which the last completed checkpoint becomes stale
values: integer >= checkpointIntervalMs
health.peerAttestationTimeoutMs5000Bound for one peer attestation call
Bound for one peer attestation call
values: integer >= 1
health.repairBacklogGraceCycles0Consecutive complete unsuccessful repair cycles with backlog during which repairFresh remains true
Consecutive complete unsuccessful repair cycles with backlog during which repairFresh remains true
values: integer >= 0
health.repairMaxAgeMs3600000Maximum age of a completed full repair sweep
Maximum age of a completed full repair sweep
values: integer >= checkpointIntervalMs
health.requiredPeerCount1 when more than one node is configured, otherwise 0Configured remote peers that must attest a round
Configured remote peers that must attest a round
values: integer >= 0
health.storageMaxAgeMscheckpointIntervalMs * 2Maximum age of the cached storage scan a checkpoint accepts
Maximum age of the cached storage scan a checkpoint accepts
values: integer >= checkpointIntervalMs
IPFS_NODE_CONFIGEnvironment variable to select config file (config.<name>.json5)
Environment variable to select config file (config.<name>.json5)
values: string
logLevelOne of fatal, error, warn, info, debug, trace, silent
One of fatal, error, warn, info, debug, trace, silent
values: fatal|error|warn|info|debug|trace|silent
maxFileCountMaximum number of files in one upload request
Maximum number of files in one upload request
values: integer 1-100
nodesPeer nodes this node replicates with and keeps connections to. May be empty for a standalone node.
Peer nodes this node replicates with and keeps connections to. May be empty for a standalone node.
values: array of objects with name and multiAddr
peerDiscovery.bootstrapMultiaddrs dialled once at startup. May be empty.
Multiaddrs dialled once at startup. May be empty.
values: array of strings
peerDiscovery.listenlibp2p listen multiaddrs, at least one
libp2p listen multiaddrs, at least one
values: array of strings
peeringSchedule*/30 * * * * *How often connected entries of nodes are pinged and missing entries are redialled
How often connected entries of nodes are pinged and missing entries are redialled
values: cron string
prettyLogsfalseRoutes logs through pino-pretty. Leave false in production
Routes logs through pino-pretty. Leave false in production
values: true|false
rateLimits.pin{ windowMs: 900000, limit: 10 }Applies to POST /api/helia/pin/:cid
Applies to POST /api/helia/pin/:cid
values: policy object
rateLimits.read{ windowMs: 60000, limit: 100 }Applies to public read routes and administrative read routes
Applies to public read routes and administrative read routes
values: policy object
rateLimits.upload{ windowMs: 900000, limit: 10 }Applies to POST /api/file/upload
Applies to POST /api/file/upload
values: policy object
replication.ackQuorum1Acknowledgements counted before an upload is reported durable
Acknowledgements counted before an upload is reported durable
values: integer >= 1
replication.enabledtrueWhether this node places copies of its own content on its peers
Whether this node places copies of its own content on its peers
values: true|false
replication.placement[{ minAgeMs: 0, copies: 4 }, { minAgeMs: 15552000000, copies: 3 }, { minAgeMs: 31536000000, copies: 2 }]Target copy count by file age, including this node
Target copy count by file age, including this node
values: array of tiers
replication.repairBatchDelayMs1000Pause between bounded passes inside one full repair cycle
Pause between bounded passes inside one full repair cycle
values: integer >= 0
replication.repairEnabledtrueRuns the scheduled repair job
Runs the scheduled repair job
values: true|false
replication.repairProbeConcurrency4Released records probed concurrently during one pass
Released records probed concurrently during one pass
values: integer >= 1
replication.repairSchedule0 */30 * * * *Repair schedule
Repair schedule
values: cron string
replication.requestTimeoutMs30000Timeout of one replication request to a peer
Timeout of one replication request to a peer
values: integer >= 1
replication.requireQuorumOnUploadfalseFails an upload with 503 when the quorum is not reached
Fails an upload with 503 when the quorum is not reached
values: true|false
serverPortHTTP API port
HTTP API port
values: integer >= 1
storage.confirmationRequiredfalseKeeps an upload temporary until POST /api/file/:cid/confirm
Keeps an upload temporary until POST /api/file/:cid/confirm
values: true|false
storage.diskReserveBytes5368709120Free space on the blockstore filesystem that uploads must never consume
Free space on the blockstore filesystem that uploads must never consume
values: integer >= 0
storage.gc.enabledtrueRuns the scheduled collector
Runs the scheduled collector
values: true|false
storage.gc.highWatermarkBytes53687091200Blockstore size above which collection starts
Blockstore size above which collection starts
values: integer >= 1
storage.gc.lowWatermarkBytes42949672960Blockstore size collection reclaims down to
Blockstore size collection reclaims down to
values: integer >= 1
storage.gc.schedule0 */15 * * * *Collector schedule
Collector schedule
values: cron string
storage.maxConcurrentDownloads64File responses retrieving content at once
File responses retrieving content at once
values: integer >= 1
storage.maxConcurrentDownloadsPerClient8Share of the global download slots one client address may hold
Share of the global download slots one client address may hold
values: integer >= 1
storage.maxConcurrentUploads32Upload requests writing into the blockstore at once
Upload requests writing into the blockstore at once
values: integer >= 1
storage.maxRequestSizeBytes536870912Combined size of all files in one upload request
Combined size of all files in one upload request
values: integer >= 1
storage.temporaryTtlMs86400000Lifetime of an unconfirmed upload before it becomes reclaimable
Lifetime of an unconfirmed upload before it becomes reclaimable
values: integer >= 1
storeFolderDirectory name resolved against the home directory of the process user
Directory name resolved against the home directory of the process user
values: string
trustProxyfalseVerified proxy hop count, or exact proxy addresses and CIDR ranges
Verified proxy hop count, or exact proxy addresses and CIDR ranges
values: false|integer >= 1|string|array of strings
uploadLimitSizeBytesMaximum size of one uploaded file
Maximum size of one uploaded file
values: integer >= 1