* filer: add StreamMutateEntry bidi streaming RPC Add a bidirectional streaming RPC that carries all filer mutation types (create, update, delete, rename) over a single ordered stream. This eliminates per-request connection overhead for pipelined operations and guarantees mutation ordering within a stream. The server handler delegates each request to the existing unary handlers (CreateEntry, UpdateEntry, DeleteEntry) and uses a proxy stream adapter for rename operations to reuse StreamRenameEntry logic. The is_last field signals completion for multi-response operations (rename sends multiple events per request; create/update/delete always send exactly one response with is_last=true). * mount: add streaming mutation multiplexer (streamMutateMux) Implement a client-side multiplexer that routes all filer mutation RPCs (create, update, delete, rename) over a single bidirectional gRPC stream. Multiple goroutines submit requests through a send channel; a dedicated sendLoop serializes them on the stream; a recvLoop dispatches responses to waiting callers via per-request channels. Key features: - Lazy stream opening on first use - Automatic reconnection on stream failure - Permanent fallback to unary RPCs if filer returns Unimplemented - Monotonic request_id for response correlation - Multi-response support for rename operations (is_last signaling) The mux is initialized on WFS and closed during unmount cleanup. No call sites use it yet — wiring comes in subsequent commits. * mount: route CreateEntry and UpdateEntry through streaming mux Wire all CreateEntry call sites to use wfs.streamCreateEntry() which routes through the StreamMutateEntry stream when available, falling back to unary RPCs otherwise. Also wire Link's UpdateEntry calls through wfs.streamUpdateEntry(). Updated call sites: - flushMetadataToFiler (file flush after write) - Mkdir (directory creation) - Symlink (symbolic link creation) - createRegularFile non-deferred path (Mknod) - flushFileMetadata (periodic metadata flush) - Link (hard link: update source + create link + rollback) * mount: route UpdateEntry and DeleteEntry through streaming mux Wire remaining mutation call sites through the streaming mux: - saveEntry (Setattr/chmod/chown/utimes) → streamUpdateEntry - Unlink → streamDeleteEntry (replaces RemoveWithResponse) - Rmdir → streamDeleteEntry (replaces RemoveWithResponse) All filer mutations except Rename now go through StreamMutateEntry when the filer supports it, with automatic unary RPC fallback. * mount: route Rename through streaming mux Wire Rename to use streamMutate.Rename() when available, with fallback to the existing StreamRenameEntry unary stream. The streaming mux sends rename as a StreamRenameEntryRequest oneof variant. The server processes it through the existing rename logic and sends multiple StreamRenameEntryResponse events (one per moved entry), with is_last=true on the final response. All filer mutations now go through a single ordered stream. * mount: fix stream mux connection ownership WithGrpcClient(streamingMode=true) closes the gRPC connection when the callback returns, destroying the stream. Own the connection directly via pb.GrpcDial so it stays alive for the stream's lifetime. Close it explicitly in recvLoop on stream failure and in Close on shutdown. * mount: fix rename failure for deferred-create files Three fixes for rename operations over the streaming mux: 1. lookupEntry: fall back to local metadata store when filer returns "not found" for entries in uncached directories. Files created with deferFilerCreate=true exist only in the local leveldb store until flushed; lookupEntry skipped the local store when the parent directory had never been readdir'd, causing rename to fail with ENOENT. 2. Rename: wait for pending async flushes and force synchronous flush of dirty metadata before sending rename to the filer. Covers the writebackCache case where close() defers the flush to a background worker that may not complete before rename fires. 3. StreamMutateEntry: propagate rename errors from server to client. Add error/errno fields to StreamMutateEntryResponse so the mount can map filer errors to correct FUSE status codes instead of silently returning OK. Also fix the existing Rename error handler which could return fuse.OK on unrecognized errors. * mount: fix streaming mux error handling, sendLoop lifecycle, and fallback Address PR review comments: 1. Server: populate top-level Error/Errno on StreamMutateEntryResponse for create/update/delete errors, not just rename. Previously update errors were silently dropped and create/delete errors were only in nested response fields that the client didn't check. 2. Client: check nested error fields in CreateEntry (ErrorCode, Error) and DeleteEntry (Error) responses, matching CreateEntryWithResponse behavior. 3. Fix sendLoop lifecycle: give each stream generation a stopSend channel. recvLoop closes it on error to stop the paired sendLoop. Previously a reconnect left the old sendLoop draining sendCh, breaking ordering. 4. Transparent fallback: stream helpers and doRename fall back to unary RPCs on transport errors (ErrStreamTransport), including the first Unimplemented from ensureStream. Previously the first call failed instead of degrading. 5. Filer rotation in openStream: try all filer addresses on dial failure, matching WithFilerClient behavior. Stop early on Unimplemented. 6. Pass metadata-bearing context to StreamMutateEntry RPC call so sw-client-id header is actually sent. 7. Gate lookupEntry local-cache fallback on open dirty handle or pending async flush to avoid resurrecting deleted/renamed entries. 8. Remove dead code in flushFileMetadata (err=nil followed by if err!=nil). 9. Use string matching for rename error-to-errno mapping in the mount to stay portable across Linux/macOS (numeric errno values differ). * mount: make failAllPending idempotent with delete-before-close Change failAllPending to collect pending entries into a local slice (deleting from the sync.Map first) before closing channels. This prevents double-close panics if called concurrently. Also remove the unused err parameter. * mount: add stream generation tracking and teardownStream Introduce a generation counter on streamMutateMux that increments each time a new stream is created. Requests carry the generation they were enqueued for so sendLoop can reject stale requests after reconnect. Add teardownStream(gen) which is idempotent (only acts when gen matches current generation and stream is non-nil). Both sendLoop and recvLoop call it on error, replacing the inline cleanup in recvLoop. sendLoop now actively triggers teardown on send errors instead of silently exiting. ensureStream waits for the prior generation's recvDone before creating a new stream, ensuring all old pending waiters are failed before reconnect. recvLoop now takes the stream, generation, and recvDone channel as parameters to avoid accessing shared fields without the lock. * mount: harden Close to prevent races with teardownStream Nil out stream, cancel, and grpcConn under the lock so that any concurrent teardownStream call from recvLoop/sendLoop becomes a no-op. Call failAllPending before closing sendCh to unblock waiters promptly. Guard recvDone with a nil check for the case where Close is called before any stream was ever opened. * mount: make errCh receive ctx-aware in doUnary and Rename Replace the blocking <-sendReq.errCh with a select that also observes ctx.Done(). If sendLoop exits via stopSend without consuming a buffered request, the caller now returns ctx.Err() instead of blocking forever. The buffered errCh (capacity 1) ensures late acknowledgements from sendLoop don't block the sender. * mount: fix sendLoop/Close race and recvLoop/teardown pending channel race Three related fixes: 1. Stop closing sendCh in Close(). Closing the shared producer channel races with callers who passed ensureStream() but haven't sent yet, causing send-on-closed-channel panics. sendCh is now left open; ensureStream checks m.closed to reject new callers. 2. Drain buffered sendCh items on shutdown. sendLoop defers drainSendCh() on exit so buffered requests get an ErrStreamTransport on their errCh instead of blocking forever. Close() drains again for any stragglers enqueued between sendLoop's drain and the final shutdown. 3. Move failAllPending from teardownStream into recvLoop's defer. teardownStream (called from sendLoop on send error) was closing pending response channels while recvLoop could be between pending.Load and the channel send — a send-on-closed-channel panic. recvLoop is now the sole closer of pending channels, eliminating the race. Close() waits on recvDone (with cancel() to guarantee Recv unblocks) so pending cleanup always completes. * filer/mount: add debug logging for hardlink lifecycle Add V(0) logging at every point where a HardLinkId is created, stored, read, or deleted to trace orphaned hardlink references. Logging covers: - gRPC server: CreateEntry/UpdateEntry when request carries HardLinkId - FilerStoreWrapper: InsertEntry/UpdateEntry when entry has HardLinkId - handleUpdateToHardLinks: entry path, HardLinkId, counter, chunk count - setHardLink: KvPut with blob size - maybeReadHardLink: V(1) on read attempt and successful decode - DeleteHardLink: counter decrement/deletion events - Mount Link(): when NewHardLinkId is generated and link is created This helps diagnose how a git pack .rev file ended up with a HardLinkId during a clone (no hard links should be involved). * test: add git clone/pull integration test for FUSE mount Shell script that exercises git operations on a SeaweedFS mount: 1. Creates a bare repo on the mount 2. Clones locally, makes 3 commits, pushes to mount 3. Clones from mount bare repo into an on-mount working dir 4. Verifies clone integrity (files, content, commit hashes) 5. Pushes 2 more commits with renames and deletes 6. Checks out an older revision on the mount clone 7. Returns to branch and pulls with real changes 8. Verifies file content, renames, deletes after pull 9. Checks git log integrity and clean status 27 assertions covering file existence, content, commit hashes, file counts, renames, deletes, and git status. Run against any existing mount: bash test-git-on-mount.sh /path/to/mount * test: add git clone/pull FUSE integration test to CI suite Add TestGitOperations to the existing fuse_integration test framework. The test exercises git's full file operation surface on the mount: 1. Creates a bare repo on the mount (acts as remote) 2. Clones locally, makes 3 commits (files, bulk data, renames), pushes 3. Clones from mount bare repo into an on-mount working dir 4. Verifies clone integrity (content, commit hash, file count) 5. Pushes 2 more commits with new files, renames, and deletes 6. Checks out an older revision on the mount clone 7. Returns to branch and pulls with real fast-forward changes 8. Verifies post-pull state: content, renames, deletes, file counts 9. Checks git log integrity (5 commits) and clean status Runs automatically in the existing fuse-integration.yml CI workflow. * mount: fix permission check with uid/gid mapping The permission checks in createRegularFile() and Access() compared the caller's local uid/gid against the entry's filer-side uid/gid without applying the uid/gid mapper. With -map.uid 501:0, a directory created as uid 0 on the filer would not match the local caller uid 501, causing hasAccess() to fall through to "other" permission bits and reject write access (0755 → other has r-x, no w). Fix: map entry uid/gid from filer-space to local-space before the hasAccess() call so both sides are in the same namespace. This fixes rsync -a failing with "Permission denied" on mkstempat when using uid/gid mapping. * mount: fix Mkdir/Symlink returning filer-side uid/gid to kernel Mkdir and Symlink used `defer wfs.mapPbIdFromFilerToLocal(entry)` to restore local uid/gid, but `outputPbEntry` writes the kernel response before the function returns — so the kernel received filer-side uid/gid (e.g., 0:0). macFUSE then caches these and rejects subsequent child operations (mkdir, create) because the caller uid (501) doesn't match the directory owner (0), and "other" bits (0755 → r-x) lack write permission. Fix: replace the defer with an explicit call to mapPbIdFromFilerToLocal before outputPbEntry, so the kernel gets local uid/gid. Also add nil guards for UidGidMapper in Access and createRegularFile to prevent panics in tests that don't configure a mapper. This fixes rsync -a "Permission denied" on mkpathat for nested directories when using uid/gid mapping. * mount: fix Link outputting filer-side uid/gid to kernel, add nil guards Link had the same defer-before-outputPbEntry bug as Mkdir and Symlink: the kernel received filer-side uid/gid because the defer hadn't run yet when outputPbEntry wrote the response. Also add nil guards for UidGidMapper in Access and createRegularFile so tests without a mapper don't panic. Audit of all outputPbEntry/outputFilerEntry call sites: - Mkdir: fixed in prior commit (explicit map before output) - Symlink: fixed in prior commit (explicit map before output) - Link: fixed here (explicit map before output) - Create (existing file): entry from maybeLoadEntry (already mapped) - Create (deferred): entry has local uid/gid (never mapped to filer) - Create (non-deferred): createRegularFile defer runs before return - Mknod: createRegularFile defer runs before return - Lookup: entry from lookupEntry (already mapped) - GetAttr: entry from maybeReadEntry/maybeLoadEntry (already mapped) - readdir: entry from cache (mapIdFromFilerToLocal) or filer (mapped) - saveEntry: no kernel output - flushMetadataToFiler: no kernel output - flushFileMetadata: no kernel output * test: fix git test for same-filesystem FUSE clone When both the bare repo and working clone live on the same FUSE mount, git's local transport uses hardlinks and cross-repo stat calls that fail on FUSE. Fix: - Use --no-local on clone to disable local transport optimizations - Use reset --hard instead of checkout to stay on branch - Use fetch + reset --hard origin/<branch> instead of git pull to avoid local transport stat failures during fetch * adjust logging * test: use plain git clone/pull to exercise real FUSE behavior Remove --no-local and fetch+reset workarounds. The test should use the same git commands users run (clone, reset --hard, pull) so it reveals real FUSE issues rather than hiding them. * test: enable V(1) logging for filer/mount and collect logs on failure - Run filer and mount with -v=1 so hardlink lifecycle logs (V(0): create/delete/insert, V(1): read attempts) are captured - On test failure, automatically dump last 16KB of all process logs (master, volume, filer, mount) to test output - Copy process logs to /tmp/seaweedfs-fuse-logs/ for CI artifact upload - Update CI workflow to upload SeaweedFS process logs alongside test output * mount: clone entry for filer flush to prevent uid/gid race flushMetadataToFiler and flushFileMetadata used entry.GetEntry() which returns the file handle's live proto entry pointer, then mutated it in-place via mapPbIdFromLocalToFiler. During the gRPC call window, a concurrent Lookup (which takes entryLock.RLock but NOT fhLockTable) could observe filer-side uid/gid (e.g., 0:0) on the file handle entry and return it to the kernel. The kernel caches these attributes, so subsequent opens by the local user (uid 501) fail with EACCES. Fix: proto.Clone the entry before mapping uid/gid for the filer request. The file handle's live entry is never mutated, so concurrent Lookup always sees local uid/gid. This fixes the intermittent "Permission denied" on .git/FETCH_HEAD after the first git pull on a mount with uid/gid mapping. * mount: add debug logging for stale lock file investigation Add V(0) logging to trace the HEAD.lock recreation issue: - Create: log when O_EXCL fails (file already exists) with uid/gid/mode - completeAsyncFlush: log resolved path, saved path, dirtyMetadata, isDeleted at entry to trace whether async flush fires after rename - flushMetadataToFiler: log the dir/name/fullpath being flushed This will show whether the async flush is recreating the lock file after git renames HEAD.lock → HEAD. * mount: prevent async flush from recreating renamed .lock files When git renames HEAD.lock → HEAD, the async flush from the prior close() can run AFTER the rename and re-insert HEAD.lock into the meta cache via its CreateEntryRequest response event. The next git pull then sees HEAD.lock and fails with "File exists". Fix: add isRenamed flag on FileHandle, set by Rename before waiting for the pending async flush. The async flush checks this flag and skips the metadata flush for renamed files (same pattern as isDeleted for unlinked files). The data pages still flush normally. The Rename handler flushes deferred metadata synchronously (Case 1) before setting isRenamed, ensuring the entry exists on the filer for the rename to proceed. For already-released handles (Case 2), the entry was created by a prior flush. * mount: also mark renamed inodes via entry.Attributes.Inode fallback When GetInode fails (Forget already removed the inode mapping), the Rename handler couldn't find the pending async flush to set isRenamed. The async flush then recreated the .lock file on the filer. Fix: fall back to oldEntry.Attributes.Inode to find the pending async flush when the inode-to-path mapping is gone. Also extract MarkInodeRenamed into a method on FileHandleToInode for clarity. * mount: skip async metadata flush when saved path no longer maps to inode The isRenamed flag approach failed for refs/remotes/origin/HEAD.lock because neither GetInode nor oldEntry.Attributes.Inode could find the inode (Forget already evicted the mapping, and the entry's stored inode was 0). Add a direct check in completeAsyncFlush: before flushing metadata, verify that the saved path still maps to this inode in the inode-to-path table. If the path was renamed or removed (inode mismatch or not found), skip the metadata flush to avoid recreating a stale entry. This catches all rename cases regardless of whether the Rename handler could set the isRenamed flag. * mount: wait for pending async flush in Unlink before filer delete Unlink was deleting the filer entry first, then marking the draining async-flush handle as deleted. The async flush worker could race between these two operations and recreate the just-unlinked entry on the filer. This caused git's .lock files (e.g. refs/remotes/origin/HEAD.lock) to persist after git pull, breaking subsequent git operations. Move the isDeleted marking and add waitForPendingAsyncFlush() before the filer delete so any in-flight flush completes first. Even if the worker raced past the isDeleted check, the wait ensures it finishes before the filer delete cleans up any recreated entry. * mount: reduce async flush and metadata flush log verbosity Raise completeAsyncFlush entry log, saved-path-mismatch skip log, and flushMetadataToFiler entry log from V(0) to V(3)/V(4). These fire for every file close with writebackCache and are too noisy for normal use. * filer: reduce hardlink debug log verbosity from V(0) to V(4) HardLinkId logs in filerstore_wrapper, filerstore_hardlink, and filer_grpc_server fire on every hardlinked file operation (git pack files use hardlinks extensively) and produce excessive noise. * mount/filer: reduce noisy V(0) logs for link, rmdir, and empty folder check - weedfs_link.go: hardlink creation logs V(0) → V(4) - weedfs_dir_mkrm.go: non-empty folder rmdir error V(0) → V(1) - empty_folder_cleaner.go: "not empty" check log V(0) → V(4) * filer: handle missing hardlink KV as expected, not error A "kv: not found" on hardlink read is normal when the link blob was already cleaned up but a stale entry still references it. Log at V(1) for not-found; keep Error level for actual KV failures. * test: add waitForDir before git pull in FUSE git operations test After git reset --hard, the FUSE mount's metadata cache may need a moment to settle on slow CI. The git pull subprocess (unpack-objects) could fail to stat the working directory. Poll for up to 5s. * Update git_operations_test.go * wait * test: simplify FUSE test framework to use weed mini Replace the 4-process setup (master + volume + filer + mount) with 2 processes: "weed mini" (all-in-one) + "weed mount". This simplifies startup, reduces port allocation, and is faster on CI. * test: fix mini flag -admin → -admin.ui
SeaweedFS
Sponsor SeaweedFS via Patreon
SeaweedFS is an independent Apache-licensed open source project with its ongoing development made possible entirely thanks to the support of these awesome backers. If you'd like to grow SeaweedFS even stronger, please consider joining our sponsors on Patreon.
Your support will be really appreciated by me and other supporters!
Gold Sponsors
- Download Binaries for different platforms
- SeaweedFS on Slack
- SeaweedFS on Twitter
- SeaweedFS on Telegram
- SeaweedFS on Reddit
- SeaweedFS Mailing List
- Wiki Documentation
- SeaweedFS White Paper
- SeaweedFS Introduction Slides 2025.5
- SeaweedFS Introduction Slides 2021.5
- SeaweedFS Introduction Slides 2019.3
Table of Contents
- Quick Start
- Introduction
- Features
- Example: Using Seaweed Object Store
- Architecture
- Compared to Other File Systems
- Dev Plan
- Installation Guide
- Disk Related Topics
- Benchmark
- Enterprise
- License
Quick Start
Quick Start with weed mini
The easiest way to get started with SeaweedFS for development and testing:
- Download the latest binary from https://github.com/seaweedfs/seaweedfs/releases and unzip a single binary file
weedorweed.exe.
Example:
# remove quarantine on macOS
# xattr -d com.apple.quarantine ./weed
./weed mini -dir=/data
This single command starts a complete SeaweedFS setup with:
- Master UI: http://localhost:9333
- Volume Server: http://localhost:9340
- Filer UI: http://localhost:8888
- S3 Endpoint: http://localhost:8333
- WebDAV: http://localhost:7333
- Admin UI: http://localhost:23646
Perfect for development, testing, learning SeaweedFS, and single node deployments!
Quick Start for S3 API on Docker
docker run -p 8333:8333 chrislusf/seaweedfs server -s3
Quick Start with Single Binary
- Download the latest binary from https://github.com/seaweedfs/seaweedfs/releases and unzip a single binary file
weedorweed.exe. Or rungo install github.com/seaweedfs/seaweedfs/weed@latest. export AWS_ACCESS_KEY_ID=admin ; export AWS_SECRET_ACCESS_KEY=keyas the admin credentials to access the object store.- Run
weed server -dir=/some/data/dir -s3to start one master, one volume server, one filer, and one S3 gateway. The difference withweed miniis thatweed minican auto configure based on the single host environment, whileweed serverrequires manual configuration and are designed for production use.
Also, to increase capacity, just add more volume servers by running weed volume -dir="/some/data/dir2" -master="<master_host>:9333" -port=8081 locally, or on a different machine, or on thousands of machines. That is it!
Introduction
SeaweedFS is a simple and highly scalable distributed file system. There are two objectives:
- to store billions of files!
- to serve the files fast!
SeaweedFS started as a blob store to handle small files efficiently. Instead of managing all file metadata in a central master, the central master only manages volumes on volume servers, and these volume servers manage files and their metadata. This relieves concurrency pressure from the central master and spreads file metadata into volume servers, allowing faster file access (O(1), usually just one disk read operation).
There is only 40 bytes of disk storage overhead for each file's metadata. It is so simple with O(1) disk reads that you are welcome to challenge the performance with your actual use cases.
SeaweedFS started by implementing Facebook's Haystack design paper. Also, SeaweedFS implements erasure coding with ideas from f4: Facebook’s Warm BLOB Storage System, and has a lot of similarities with Facebook’s Tectonic Filesystem and Google's Colossus File System
On top of the blob store, optional Filer can support directories and POSIX attributes. Filer is a separate linearly-scalable stateless server with customizable metadata stores, e.g., MySql, Postgres, Redis, Cassandra, HBase, Mongodb, Elastic Search, LevelDB, RocksDB, Sqlite, MemSql, TiDB, Etcd, CockroachDB, YDB, etc.
SeaweedFS can transparently integrate with the cloud. With hot data on local cluster, and warm data on the cloud with O(1) access time, SeaweedFS can achieve both fast local access time and elastic cloud storage capacity. What's more, the cloud storage access API cost is minimized. Faster and cheaper than direct cloud storage!
Features
Additional Blob Store Features
- Support different replication levels, with rack and data center aware.
- Automatic master servers failover - no single point of failure (SPOF).
- Automatic compression depending on file MIME type.
- Automatic compaction to reclaim disk space after deletion or update.
- Automatic entry TTL expiration.
- Flexible Capacity Expansion: Any server with some disk space can add to the total storage space.
- Adding/Removing servers does not cause any data re-balancing unless triggered by admin commands.
- Optional picture resizing.
- Support ETag, Accept-Range, Last-Modified, etc.
- Support in-memory/leveldb/readonly mode tuning for memory/performance balance.
- Support rebalancing the writable and readonly volumes.
- Customizable Multiple Storage Tiers: Customizable storage disk types to balance performance and cost.
- Transparent cloud integration: unlimited capacity via tiered cloud storage for warm data.
- Erasure Coding for warm storage Rack-Aware 10.4 erasure coding reduces storage cost and increases availability. Enterprise version can customize EC ratio.
Filer Features
- Filer server provides "normal" directories and files via HTTP.
- File TTL automatically expires file metadata and actual file data.
- Mount filer reads and writes files directly as a local directory via FUSE.
- Filer Store Replication enables HA for filer meta data stores.
- Active-Active Replication enables asynchronous one-way or two-way cross cluster continuous replication.
- Amazon S3 compatible API accesses files with S3 tooling.
- Hadoop Compatible File System accesses files from Hadoop/Spark/Flink/etc or even runs HBase.
- Async Replication To Cloud has extremely fast local access and backups to Amazon S3, Google Cloud Storage, Azure, BackBlaze.
- WebDAV accesses as a mapped drive on Mac and Windows, or from mobile devices.
- AES256-GCM Encrypted Storage safely stores the encrypted data.
- Super Large Files stores large or super large files in tens of TB.
- Cloud Drive mounts cloud storage to local cluster, cached for fast read and write with asynchronous write back.
- Gateway to Remote Object Store mirrors bucket operations to remote object storage, in addition to Cloud Drive
Kubernetes
- Kubernetes CSI Driver A Container Storage Interface (CSI) Driver.
- SeaweedFS Operator
Example: Using Seaweed Blob Store
By default, the master node runs on port 9333, and the volume nodes run on port 8080. Let's start one master node, and two volume nodes on port 8080 and 8081. Ideally, they should be started from different machines. We'll use localhost as an example.
SeaweedFS uses HTTP REST operations to read, write, and delete. The responses are in JSON or JSONP format.
Start Master Server
> ./weed master
Start Volume Servers
> weed volume -dir="/tmp/data1" -max=5 -master="localhost:9333" -port=8080 &
> weed volume -dir="/tmp/data2" -max=10 -master="localhost:9333" -port=8081 &
Write A Blob
A blob, also referred as a needle, a chunk, or mistakenly as a file, is just a byte array. It can have attributes, such as name, mime type, create or update time, etc. But basically it is just a byte array of a relatively small size, such as 2 MB ~ 64 MB. The size is not fixed.
To upload a blob: first, send a HTTP POST, PUT, or GET request to /dir/assign to get an fid and a volume server URL:
> curl http://localhost:9333/dir/assign
{"count":1,"fid":"3,01637037d6","url":"127.0.0.1:8080","publicUrl":"localhost:8080"}
Second, to store the blob content, send a HTTP multi-part POST request to url + '/' + fid from the response:
> curl -F file=@/home/chris/myphoto.jpg http://127.0.0.1:8080/3,01637037d6
{"name":"myphoto.jpg","size":43234,"eTag":"1cc0118e"}
To update, send another POST request with updated blob content.
For deletion, send an HTTP DELETE request to the same url + '/' + fid URL:
> curl -X DELETE http://127.0.0.1:8080/3,01637037d6
Save Blob Id
Now, you can save the fid, 3,01637037d6 in this case, to a database field.
The number 3 at the start represents a volume id. After the comma, it's one file key, 01, and a file cookie, 637037d6.
The volume id is an unsigned 32-bit integer. The file key is an unsigned 64-bit integer. The file cookie is an unsigned 32-bit integer, used to prevent URL guessing.
The file key and file cookie are both coded in hex. You can store the <volume id, file key, file cookie> tuple in your own format, or simply store the fid as a string.
If stored as a string, in theory, you would need 8+1+16+8=33 bytes. A char(33) would be enough, if not more than enough, since most uses will not need 2^32 volumes.
If space is really a concern, you can store the file id in the binary format. You would need one 4-byte integer for volume id, 8-byte long number for file key, and a 4-byte integer for the file cookie. So 16 bytes are more than enough.
Read a Blob
Here is an example of how to render the URL.
First look up the volume server's URLs by the file's volumeId:
> curl http://localhost:9333/dir/lookup?volumeId=3
{"volumeId":"3","locations":[{"publicUrl":"localhost:8080","url":"localhost:8080"}]}
Since (usually) there are not too many volume servers, and volumes don't move often, you can cache the results most of the time. Depending on the replication type, one volume can have multiple replica locations. Just randomly pick one location to read.
Now you can take the public URL, render the URL or directly read from the volume server via URL:
http://localhost:8080/3,01637037d6.jpg
Notice we add a file extension ".jpg" here. It's optional and just one way for the client to specify the file content type.
If you want a nicer URL, you can use one of these alternative URL formats:
http://localhost:8080/3/01637037d6/my_preferred_name.jpg
http://localhost:8080/3/01637037d6.jpg
http://localhost:8080/3,01637037d6.jpg
http://localhost:8080/3/01637037d6
http://localhost:8080/3,01637037d6
If you want to get a scaled version of an image, you can add some params:
http://localhost:8080/3/01637037d6.jpg?height=200&width=200
http://localhost:8080/3/01637037d6.jpg?height=200&width=200&mode=fit
http://localhost:8080/3/01637037d6.jpg?height=200&width=200&mode=fill
Rack-Aware and Data Center-Aware Replication
SeaweedFS applies the replication strategy at a volume level. So, when you are getting a blob id, you can specify the replication strategy. For example:
curl http://localhost:9333/dir/assign?replication=001
The replication parameter options are:
000: no replication
001: replicate once on the same rack
010: replicate once on a different rack, but same data center
100: replicate once on a different data center
200: replicate twice on two different data center
110: replicate once on a different rack, and once on a different data center
More details about replication can be found on the wiki.
You can also set the default replication strategy when starting the master server.
Allocate Blob Key on Specific Data Center
Volume servers can be started with a specific data center name:
weed volume -dir=/tmp/1 -port=8080 -dataCenter=dc1
weed volume -dir=/tmp/2 -port=8081 -dataCenter=dc2
When requesting a blob key, an optional "dataCenter" parameter can limit the assigned volume to the specific data center. For example, this specifies that the assigned volume should be limited to 'dc1':
http://localhost:9333/dir/assign?dataCenter=dc1
Other Features
- No Single Point of Failure
- Insert with your own keys
- Chunking large files
- Collection as a Simple Name Space
Blob Store Architecture
Usually distributed file systems split each file into chunks. A central server keeps a mapping of filenames to chunks, and also which chunks each chunk server has.
The main drawback is that the central server can't handle many small files efficiently, and since all read requests need to go through the central master, so it might not scale well for many concurrent users.
Instead of managing chunks, SeaweedFS manages data volumes in the master server. Each data volume is 32GB in size, and can hold a lot of blobs. And each storage node can have many data volumes. So the master node only needs to store the metadata about the volumes, which is a fairly small amount of data and is generally stable.
The actual blob metadata, which are the blob volume, offset, and size, is stored in each volume on volume servers. Since each volume server only manages metadata of blobs on its own disk, with only 16 bytes for each blob, all access can read the metadata just from memory and only needs one disk operation to actually read file data.
For comparison, consider that an xfs inode structure in Linux is 536 bytes.
Master Server and Volume Server
The architecture is fairly simple. The actual data is stored in volumes on storage nodes. One volume server can have multiple volumes, and can both support read and write access with basic authentication.
All volumes are managed by a master server. The master server contains the volume id to volume server mapping. This is fairly static information, and can be easily cached.
On each write request, the master server also generates a file key, which is a growing 64-bit unsigned integer. Since write requests are not generally as frequent as read requests, one master server should be able to handle the concurrency well.
Write and Read files
When a client sends a write request, the master server returns (volume id, file key, file cookie, volume node URL) for the blob. The client then contacts the volume node and POSTs the blob content.
When a client needs to read a blob based on (volume id, file key, file cookie), it asks the master server by the volume id for the (volume node URL, volume node public URL), or retrieves this from a cache. Then the client can GET the content, or just render the URL on web pages and let browsers fetch the content.
Saving memory
All blob metadata stored on a volume server is readable from memory without disk access. Each file takes just a 16-byte map entry of <64bit key, 32bit offset, 32bit size>. Of course, each map entry has its own space cost for the map. But usually the disk space runs out before the memory does.
Tiered Storage to the cloud
The local volume servers are much faster, while cloud storages have elastic capacity and are actually more cost-efficient if not accessed often (usually free to upload, but relatively costly to access). With the append-only structure and O(1) access time, SeaweedFS can take advantage of both local and cloud storage by offloading the warm data to the cloud.
Usually hot data are fresh and warm data are old. SeaweedFS puts the newly created volumes on local servers, and optionally upload the older volumes on the cloud. If the older data are accessed less often, this literally gives you unlimited capacity with limited local servers, and still fast for new data.
With the O(1) access time, the network latency cost is kept at minimum.
If the hot/warm data is split as 20/80, with 20 servers, you can achieve storage capacity of 100 servers. That's a cost saving of 80%! Or you can repurpose the 80 servers to store new data also, and get 5X storage throughput.
SeaweedFS Filer
Built on top of the blob store, SeaweedFS Filer adds directory structure to create a file system. The directory sturcture is an interface that is implemented in many key-value stores or databases.
The content of a file is mapped to one or many blobs, distributed to multiple volumes on multiple volume servers.
Compared to Other File Systems
Most other distributed file systems seem more complicated than necessary.
SeaweedFS is meant to be fast and simple, in both setup and operation. If you do not understand how it works when you reach here, we've failed! Please raise an issue with any questions or update this file with clarifications.
SeaweedFS is constantly moving forward. Same with other systems. These comparisons can be outdated quickly. Please help to keep them updated.
Compared to HDFS
HDFS uses the chunk approach for each file, and is ideal for storing large files.
SeaweedFS is ideal for serving relatively smaller files quickly and concurrently.
SeaweedFS can also store extra large files by splitting them into manageable data chunks, and store the file ids of the data chunks into a meta chunk. This is managed by "weed upload/download" tool, and the weed master or volume servers are agnostic about it.
Compared to GlusterFS, Ceph
The architectures are mostly the same. SeaweedFS aims to store and read files fast, with a simple and flat architecture. The main differences are
- SeaweedFS optimizes for small files, ensuring O(1) disk seek operation, and can also handle large files.
- SeaweedFS statically assigns a volume id for a file. Locating file content becomes just a lookup of the volume id, which can be easily cached.
- SeaweedFS Filer metadata store can be any well-known and proven data store, e.g., Redis, Cassandra, HBase, Mongodb, Elastic Search, MySql, Postgres, Sqlite, MemSql, TiDB, CockroachDB, Etcd, YDB etc, and is easy to customize.
- SeaweedFS Volume server also communicates directly with clients via HTTP, supporting range queries, direct uploads, etc.
| System | File Metadata | File Content Read | POSIX | REST API | Optimized for large number of small files |
|---|---|---|---|---|---|
| SeaweedFS | lookup volume id, cacheable | O(1) disk seek | Yes | Yes | |
| SeaweedFS Filer | Linearly Scalable, Customizable | O(1) disk seek | FUSE | Yes | Yes |
| GlusterFS | hashing | FUSE, NFS | |||
| Ceph | hashing + rules | FUSE | Yes | ||
| MooseFS | in memory | FUSE | No | ||
| MinIO | separate meta file for each file | Yes | No |
Compared to GlusterFS
GlusterFS stores files, both directories and content, in configurable volumes called "bricks".
GlusterFS hashes the path and filename into ids, and assigned to virtual volumes, and then mapped to "bricks".
Compared to MooseFS
MooseFS chooses to neglect small file issue. From moosefs 3.0 manual, "even a small file will occupy 64KiB plus additionally 4KiB of checksums and 1KiB for the header", because it "was initially designed for keeping large amounts (like several thousands) of very big files"
MooseFS Master Server keeps all meta data in memory. Same issue as HDFS namenode.
Compared to Ceph
Ceph can be setup similar to SeaweedFS as a key->blob store. It is much more complicated, with the need to support layers on top of it. Here is a more detailed comparison
SeaweedFS has a centralized master group to look up free volumes, while Ceph uses hashing and metadata servers to locate its objects. Having a centralized master makes it easy to code and manage.
Ceph, like SeaweedFS, is based on the object store RADOS. Ceph is rather complicated with mixed reviews.
Ceph uses CRUSH hashing to automatically manage data placement, which is efficient to locate the data. But the data has to be placed according to the CRUSH algorithm. Any wrong configuration would cause data loss. Topology changes, such as adding new servers to increase capacity, will cause data migration with high IO cost to fit the CRUSH algorithm. SeaweedFS places data by assigning them to any writable volumes. If writes to one volume failed, just pick another volume to write. Adding more volumes is also as simple as it can be.
SeaweedFS is optimized for small files. Small files are stored as one continuous block of content, with at most 8 unused bytes between files. Small file access is O(1) disk read.
SeaweedFS Filer uses off-the-shelf stores, such as MySql, Postgres, Sqlite, Mongodb, Redis, Elastic Search, Cassandra, HBase, MemSql, TiDB, CockroachCB, Etcd, YDB, to manage file directories. These stores are proven, scalable, and easier to manage.
| SeaweedFS | comparable to Ceph | advantage |
|---|---|---|
| Master | MDS | simpler |
| Volume | OSD | optimized for small files |
| Filer | Ceph FS | linearly scalable, Customizable, O(1) or O(logN) |
Compared to MinIO
MinIO follows AWS S3 closely and is ideal for testing for S3 API. It has good UI, policies, versionings, etc. SeaweedFS is trying to catch up here. It is also possible to put MinIO as a gateway in front of SeaweedFS later.
MinIO metadata are in simple files. Each file write will incur extra writes to corresponding meta file.
MinIO does not have optimization for lots of small files. The files are simply stored as is to local disks. Plus the extra meta file and shards for erasure coding, it only amplifies the LOSF problem.
MinIO has multiple disk IO to read one file. SeaweedFS has O(1) disk reads, even for erasure coded files.
MinIO has full-time erasure coding. SeaweedFS uses replication on hot data for faster speed and optionally applies erasure coding on warm data.
MinIO does not have POSIX-like API support.
MinIO has specific requirements on storage layout. It is not flexible to adjust capacity. In SeaweedFS, just start one volume server pointing to the master. That's all.
Dev Plan
- More tools and documentation, on how to manage and scale the system.
- Read and write stream data.
- Support structured data.
This is a super exciting project! And we need helpers and support!
Installation Guide
Installation guide for users who are not familiar with golang
Step 1: install go on your machine and setup the environment by following the instructions at:
https://golang.org/doc/install
make sure to define your $GOPATH
Step 2: checkout this repo:
git clone https://github.com/seaweedfs/seaweedfs.git
Step 3: download, compile, and install the project by executing the following command
cd seaweedfs/weed && make install
Once this is done, you will find the executable "weed" in your $GOPATH/bin directory
For more installation options, including how to run with Docker, see the Getting Started guide.
Disk Related Topics
Hard Drive Performance
When testing read performance on SeaweedFS, it basically becomes a performance test of your hard drive's random read speed. Hard drives usually get 100MB/s~200MB/s.
Solid State Disk
To modify or delete small files, SSD must delete a whole block at a time, and move content in existing blocks to a new block. SSD is fast when brand new, but will get fragmented over time and you have to garbage collect, compacting blocks. SeaweedFS is friendly to SSD since it is append-only. Deletion and compaction are done on volume level in the background, not slowing reading and not causing fragmentation.
Benchmark
My Own Unscientific Single Machine Results on Mac Book with Solid State Disk, CPU: 1 Intel Core i7 2.6GHz.
Write 1 million 1KB file:
Concurrency Level: 16
Time taken for tests: 66.753 seconds
Completed requests: 1048576
Failed requests: 0
Total transferred: 1106789009 bytes
Requests per second: 15708.23 [#/sec]
Transfer rate: 16191.69 [Kbytes/sec]
Connection Times (ms)
min avg max std
Total: 0.3 1.0 84.3 0.9
Percentage of the requests served within a certain time (ms)
50% 0.8 ms
66% 1.0 ms
75% 1.1 ms
80% 1.2 ms
90% 1.4 ms
95% 1.7 ms
98% 2.1 ms
99% 2.6 ms
100% 84.3 ms
Randomly read 1 million files:
Concurrency Level: 16
Time taken for tests: 22.301 seconds
Completed requests: 1048576
Failed requests: 0
Total transferred: 1106812873 bytes
Requests per second: 47019.38 [#/sec]
Transfer rate: 48467.57 [Kbytes/sec]
Connection Times (ms)
min avg max std
Total: 0.0 0.3 54.1 0.2
Percentage of the requests served within a certain time (ms)
50% 0.3 ms
90% 0.4 ms
98% 0.6 ms
99% 0.7 ms
100% 54.1 ms
Run WARP and launch a mixed benchmark.
make benchmark
warp: Benchmark data written to "warp-mixed-2025-12-05[194844]-kBpU.csv.zst"
Mixed operations.
Operation: DELETE, 10%, Concurrency: 20, Ran 42s.
* Throughput: 55.13 obj/s
Operation: GET, 45%, Concurrency: 20, Ran 42s.
* Throughput: 2477.45 MiB/s, 247.75 obj/s
Operation: PUT, 15%, Concurrency: 20, Ran 42s.
* Throughput: 825.85 MiB/s, 82.59 obj/s
Operation: STAT, 30%, Concurrency: 20, Ran 42s.
* Throughput: 165.27 obj/s
Cluster Total: 3302.88 MiB/s, 550.51 obj/s over 43s.
Enterprise
For enterprise users, please visit seaweedfs.com for the SeaweedFS Enterprise Edition, which has a self-healing storage format with better data protection.
License
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
The text of this page is available for modification and reuse under the terms of the Creative Commons Attribution-Sharealike 3.0 Unported License and the GNU Free Documentation License (unversioned, with no invariant sections, front-cover texts, or back-cover texts).



