--- title: "rclone serve s3" description: "Serve remote:path over s3." slug: rclone_serve_s3 url: /commands/rclone_serve_s3/ groups: Filter status: Experimental versionIntroduced: v1.65 # autogenerated - DO NOT EDIT, instead edit the source code in cmd/serve/s3/ and as part of making a release run "make commanddocs" --- # rclone serve s3 Serve remote:path over s3. ## Synopsis `serve s3` implements a basic s3 server that serves a remote via s3. This can be viewed with an s3 client, or you can make an [s3 type remote](/s3/) to read and write to it with rclone. `serve s3` is considered **Experimental** so use with care. S3 server supports Signature Version 4 authentication. Just use `--auth-key accessKey,secretKey` and set the `Authorization` header correctly in the request. (See the [AWS docs](https://docs.aws.amazon.com/general/latest/gr/signature-version-4.html)). `--auth-key` can be repeated for multiple auth pairs. If `--auth-key` is not provided then `serve s3` will allow anonymous access. Please note that some clients may require HTTPS endpoints. See [the SSL docs](#ssl-tls) for more information. This command uses the [VFS directory cache](#vfs-virtual-file-system). All the functionality will work with `--vfs-cache-mode off`. Using `--vfs-cache-mode full` (or `writes`) can be used to cache objects locally to improve performance. Use `--force-path-style=false` if you want to use the bucket name as a part of the hostname (such as mybucket.local) Use `--etag-hash` if you want to change the hash uses for the `ETag`. Note that using anything other than `MD5` (the default) is likely to cause problems for S3 clients which rely on the Etag being the MD5. ## Quickstart For a simple set up, to serve `remote:path` over s3, run the server like this: ``` rclone serve s3 --auth-key ACCESS_KEY_ID,SECRET_ACCESS_KEY remote:path ``` This will be compatible with an rclone remote which is defined like this: ``` [serves3] type = s3 provider = Rclone endpoint = http://127.0.0.1:8080/ access_key_id = ACCESS_KEY_ID secret_access_key = SECRET_ACCESS_KEY use_multipart_uploads = false ``` Note that setting `disable_multipart_uploads = true` is to work around [a bug](#bugs) which will be fixed in due course. ## Bugs When uploading multipart files `serve s3` holds all the parts in memory (see [#7453](https://github.com/rclone/rclone/issues/7453)). This is a limitaton of the library rclone uses for serving S3 and will hopefully be fixed at some point. Multipart server side copies do not work (see [#7454](https://github.com/rclone/rclone/issues/7454)). These take a very long time and eventually fail. The default threshold for multipart server side copies is 5G which is the maximum it can be, so files above this side will fail to be server side copied. For a current list of `serve s3` bugs see the [serve s3](https://github.com/rclone/rclone/labels/serve%20s3) bug category on GitHub. ## Limitations `serve s3` will treat all directories in the root as buckets and ignore all files in the root. You can use `CreateBucket` to create folders under the root, but you can't create empty folders under other folders not in the root. When using `PutObject` or `DeleteObject`, rclone will automatically create or clean up empty folders. If you don't want to clean up empty folders automatically, use `--no-cleanup`. When using `ListObjects`, rclone will use `/` when the delimiter is empty. This reduces backend requests with no effect on most operations, but if the delimiter is something other than `/` and empty, rclone will do a full recursive search of the backend, which can take some time. Versioning is not currently supported. Metadata will only be saved in memory other than the rclone `mtime` metadata which will be set as the modification time of the file. ## Supported operations `serve s3` currently supports the following operations. - Bucket - `ListBuckets` - `CreateBucket` - `DeleteBucket` - Object - `HeadObject` - `ListObjects` - `GetObject` - `PutObject` - `DeleteObject` - `DeleteObjects` - `CreateMultipartUpload` - `CompleteMultipartUpload` - `AbortMultipartUpload` - `CopyObject` - `UploadPart` Other operations will return error `Unimplemented`. ## Server options Use `--addr` to specify which IP address and port the server should listen on, eg `--addr 1.2.3.4:8000` or `--addr :8080` to listen to all IPs. By default it only listens on localhost. You can use port :0 to let the OS choose an available port. If you set `--addr` to listen on a public or LAN accessible IP address then using Authentication is advised - see the next section for info. You can use a unix socket by setting the url to `unix:///path/to/socket` or just by using an absolute path name. Note that unix sockets bypass the authentication - this is expected to be done with file system permissions. `--addr` may be repeated to listen on multiple IPs/ports/sockets. `--server-read-timeout` and `--server-write-timeout` can be used to control the timeouts on the server. Note that this is the total time for a transfer. `--max-header-bytes` controls the maximum number of bytes the server will accept in the HTTP header. `--baseurl` controls the URL prefix that rclone serves from. By default rclone will serve from the root. If you used `--baseurl "/rclone"` then rclone would serve from a URL starting with "/rclone/". This is useful if you wish to proxy rclone serve. Rclone automatically inserts leading and trailing "/" on `--baseurl`, so `--baseurl "rclone"`, `--baseurl "/rclone"` and `--baseurl "/rclone/"` are all treated identically. ### TLS (SSL) By default this will serve over http. If you want you can serve over https. You will need to supply the `--cert` and `--key` flags. If you wish to do client side certificate validation then you will need to supply `--client-ca` also. `--cert` should be a either a PEM encoded certificate or a concatenation of that with the CA certificate. `--key` should be the PEM encoded private key and `--client-ca` should be the PEM encoded client certificate authority certificate. --min-tls-version is minimum TLS version that is acceptable. Valid values are "tls1.0", "tls1.1", "tls1.2" and "tls1.3" (default "tls1.0"). ## VFS - Virtual File System This command uses the VFS layer. This adapts the cloud storage objects that rclone uses into something which looks much more like a disk filing system. Cloud storage objects have lots of properties which aren't like disk files - you can't extend them or write to the middle of them, so the VFS layer has to deal with that. Because there is no one right way of doing this there are various options explained below. The VFS layer also implements a directory cache - this caches info about files and directories (but not the data) in memory. ## VFS Directory Cache Using the `--dir-cache-time` flag, you can control how long a directory should be considered up to date and not refreshed from the backend. Changes made through the VFS will appear immediately or invalidate the cache. --dir-cache-time duration Time to cache directory entries for (default 5m0s) --poll-interval duration Time to wait between polling for changes. Must be smaller than dir-cache-time. Only on supported remotes. Set to 0 to disable (default 1m0s) However, changes made directly on the cloud storage by the web interface or a different copy of rclone will only be picked up once the directory cache expires if the backend configured does not support polling for changes. If the backend supports polling, changes will be picked up within the polling interval. You can send a `SIGHUP` signal to rclone for it to flush all directory caches, regardless of how old they are. Assuming only one rclone instance is running, you can reset the cache like this: kill -SIGHUP $(pidof rclone) If you configure rclone with a [remote control](/rc) then you can use rclone rc to flush the whole directory cache: rclone rc vfs/forget Or individual files or directories: rclone rc vfs/forget file=path/to/file dir=path/to/dir ## VFS File Buffering The `--buffer-size` flag determines the amount of memory, that will be used to buffer data in advance. Each open file will try to keep the specified amount of data in memory at all times. The buffered data is bound to one open file and won't be shared. This flag is a upper limit for the used memory per open file. The buffer will only use memory for data that is downloaded but not not yet read. If the buffer is empty, only a small amount of memory will be used. The maximum memory used by rclone for buffering can be up to `--buffer-size * open files`. ## VFS File Caching These flags control the VFS file caching options. File caching is necessary to make the VFS layer appear compatible with a normal file system. It can be disabled at the cost of some compatibility. For example you'll need to enable VFS caching if you want to read and write simultaneously to a file. See below for more details. Note that the VFS cache is separate from the cache backend and you may find that you need one or the other or both. --cache-dir string Directory rclone will use for caching. --vfs-cache-mode CacheMode Cache mode off|minimal|writes|full (default off) --vfs-cache-max-age duration Max time since last access of objects in the cache (default 1h0m0s) --vfs-cache-max-size SizeSuffix Max total size of objects in the cache (default off) --vfs-cache-min-free-space SizeSuffix Target minimum free space on the disk containing the cache (default off) --vfs-cache-poll-interval duration Interval to poll the cache for stale objects (default 1m0s) --vfs-write-back duration Time to writeback files after last use when using cache (default 5s) If run with `-vv` rclone will print the location of the file cache. The files are stored in the user cache file area which is OS dependent but can be controlled with `--cache-dir` or setting the appropriate environment variable. The cache has 4 different modes selected by `--vfs-cache-mode`. The higher the cache mode the more compatible rclone becomes at the cost of using disk space. Note that files are written back to the remote only when they are closed and if they haven't been accessed for `--vfs-write-back` seconds. If rclone is quit or dies with files that haven't been uploaded, these will be uploaded next time rclone is run with the same flags. If using `--vfs-cache-max-size` or `--vfs-cache-min-free-size` note that the cache may exceed these quotas for two reasons. Firstly because it is only checked every `--vfs-cache-poll-interval`. Secondly because open files cannot be evicted from the cache. When `--vfs-cache-max-size` or `--vfs-cache-min-free-size` is exceeded, rclone will attempt to evict the least accessed files from the cache first. rclone will start with files that haven't been accessed for the longest. This cache flushing strategy is efficient and more relevant files are likely to remain cached. The `--vfs-cache-max-age` will evict files from the cache after the set time since last access has passed. The default value of 1 hour will start evicting files from cache that haven't been accessed for 1 hour. When a cached file is accessed the 1 hour timer is reset to 0 and will wait for 1 more hour before evicting. Specify the time with standard notation, s, m, h, d, w . You **should not** run two copies of rclone using the same VFS cache with the same or overlapping remotes if using `--vfs-cache-mode > off`. This can potentially cause data corruption if you do. You can work around this by giving each rclone its own cache hierarchy with `--cache-dir`. You don't need to worry about this if the remotes in use don't overlap. ### --vfs-cache-mode off In this mode (the default) the cache will read directly from the remote and write directly to the remote without caching anything on disk. This will mean some operations are not possible * Files can't be opened for both read AND write * Files opened for write can't be seeked * Existing files opened for write must have O_TRUNC set * Files open for read with O_TRUNC will be opened write only * Files open for write only will behave as if O_TRUNC was supplied * Open modes O_APPEND, O_TRUNC are ignored * If an upload fails it can't be retried ### --vfs-cache-mode minimal This is very similar to "off" except that files opened for read AND write will be buffered to disk. This means that files opened for write will be a lot more compatible, but uses the minimal disk space. These operations are not possible * Files opened for write only can't be seeked * Existing files opened for write must have O_TRUNC set * Files opened for write only will ignore O_APPEND, O_TRUNC * If an upload fails it can't be retried ### --vfs-cache-mode writes In this mode files opened for read only are still read directly from the remote, write only and read/write files are buffered to disk first. This mode should support all normal file system operations. If an upload fails it will be retried at exponentially increasing intervals up to 1 minute. ### --vfs-cache-mode full In this mode all reads and writes are buffered to and from disk. When data is read from the remote this is buffered to disk as well. In this mode the files in the cache will be sparse files and rclone will keep track of which bits of the files it has downloaded. So if an application only reads the starts of each file, then rclone will only buffer the start of the file. These files will appear to be their full size in the cache, but they will be sparse files with only the data that has been downloaded present in them. This mode should support all normal file system operations and is otherwise identical to `--vfs-cache-mode` writes. When reading a file rclone will read `--buffer-size` plus `--vfs-read-ahead` bytes ahead. The `--buffer-size` is buffered in memory whereas the `--vfs-read-ahead` is buffered on disk. When using this mode it is recommended that `--buffer-size` is not set too large and `--vfs-read-ahead` is set large if required. **IMPORTANT** not all file systems support sparse files. In particular FAT/exFAT do not. Rclone will perform very badly if the cache directory is on a filesystem which doesn't support sparse files and it will log an ERROR message if one is detected. ### Fingerprinting Various parts of the VFS use fingerprinting to see if a local file copy has changed relative to a remote file. Fingerprints are made from: - size - modification time - hash where available on an object. On some backends some of these attributes are slow to read (they take an extra API call per object, or extra work per object). For example `hash` is slow with the `local` and `sftp` backends as they have to read the entire file and hash it, and `modtime` is slow with the `s3`, `swift`, `ftp` and `qinqstor` backends because they need to do an extra API call to fetch it. If you use the `--vfs-fast-fingerprint` flag then rclone will not include the slow operations in the fingerprint. This makes the fingerprinting less accurate but much faster and will improve the opening time of cached files. If you are running a vfs cache over `local`, `s3` or `swift` backends then using this flag is recommended. Note that if you change the value of this flag, the fingerprints of the files in the cache may be invalidated and the files will need to be downloaded again. ## VFS Chunked Reading When rclone reads files from a remote it reads them in chunks. This means that rather than requesting the whole file rclone reads the chunk specified. This can reduce the used download quota for some remotes by requesting only chunks from the remote that are actually read, at the cost of an increased number of requests. These flags control the chunking: --vfs-read-chunk-size SizeSuffix Read the source objects in chunks (default 128M) --vfs-read-chunk-size-limit SizeSuffix Max chunk doubling size (default off) Rclone will start reading a chunk of size `--vfs-read-chunk-size`, and then double the size for each read. When `--vfs-read-chunk-size-limit` is specified, and greater than `--vfs-read-chunk-size`, the chunk size for each open file will get doubled only until the specified value is reached. If the value is "off", which is the default, the limit is disabled and the chunk size will grow indefinitely. With `--vfs-read-chunk-size 100M` and `--vfs-read-chunk-size-limit 0` the following parts will be downloaded: 0-100M, 100M-200M, 200M-300M, 300M-400M and so on. When `--vfs-read-chunk-size-limit 500M` is specified, the result would be 0-100M, 100M-300M, 300M-700M, 700M-1200M, 1200M-1700M and so on. Setting `--vfs-read-chunk-size` to `0` or "off" disables chunked reading. ## VFS Performance These flags may be used to enable/disable features of the VFS for performance or other reasons. See also the [chunked reading](#vfs-chunked-reading) feature. In particular S3 and Swift benefit hugely from the `--no-modtime` flag (or use `--use-server-modtime` for a slightly different effect) as each read of the modification time takes a transaction. --no-checksum Don't compare checksums on up/download. --no-modtime Don't read/write the modification time (can speed things up). --no-seek Don't allow seeking in files. --read-only Only allow read-only access. Sometimes rclone is delivered reads or writes out of order. Rather than seeking rclone will wait a short time for the in sequence read or write to come in. These flags only come into effect when not using an on disk cache file. --vfs-read-wait duration Time to wait for in-sequence read before seeking (default 20ms) --vfs-write-wait duration Time to wait for in-sequence write before giving error (default 1s) When using VFS write caching (`--vfs-cache-mode` with value writes or full), the global flag `--transfers` can be set to adjust the number of parallel uploads of modified files from the cache (the related global flag `--checkers` has no effect on the VFS). --transfers int Number of file transfers to run in parallel (default 4) ## VFS Case Sensitivity Linux file systems are case-sensitive: two files can differ only by case, and the exact case must be used when opening a file. File systems in modern Windows are case-insensitive but case-preserving: although existing files can be opened using any case, the exact case used to create the file is preserved and available for programs to query. It is not allowed for two files in the same directory to differ only by case. Usually file systems on macOS are case-insensitive. It is possible to make macOS file systems case-sensitive but that is not the default. The `--vfs-case-insensitive` VFS flag controls how rclone handles these two cases. If its value is "false", rclone passes file names to the remote as-is. If the flag is "true" (or appears without a value on the command line), rclone may perform a "fixup" as explained below. The user may specify a file name to open/delete/rename/etc with a case different than what is stored on the remote. If an argument refers to an existing file with exactly the same name, then the case of the existing file on the disk will be used. However, if a file name with exactly the same name is not found but a name differing only by case exists, rclone will transparently fixup the name. This fixup happens only when an existing file is requested. Case sensitivity of file names created anew by rclone is controlled by the underlying remote. Note that case sensitivity of the operating system running rclone (the target) may differ from case sensitivity of a file system presented by rclone (the source). The flag controls whether "fixup" is performed to satisfy the target. If the flag is not provided on the command line, then its default value depends on the operating system where rclone runs: "true" on Windows and macOS, "false" otherwise. If the flag is provided without a value, then it is "true". The `--no-unicode-normalization` flag controls whether a similar "fixup" is performed for filenames that differ but are [canonically equivalent](https://en.wikipedia.org/wiki/Unicode_equivalence) with respect to unicode. Unicode normalization can be particularly helpful for users of macOS, which prefers form NFD instead of the NFC used by most other platforms. It is therefore highly recommended to keep the default of `false` on macOS, to avoid encoding compatibility issues. In the (probably unlikely) event that a directory has multiple duplicate filenames after applying case and unicode normalization, the `--vfs-block-norm-dupes` flag allows hiding these duplicates. This comes with a performance tradeoff, as rclone will have to scan the entire directory for duplicates when listing a directory. For this reason, it is recommended to leave this disabled if not needed. However, macOS users may wish to consider using it, as otherwise, if a remote directory contains both NFC and NFD versions of the same filename, an odd situation will occur: both versions of the file will be visible in the mount, and both will appear to be editable, however, editing either version will actually result in only the NFD version getting edited under the hood. `--vfs-block- norm-dupes` prevents this confusion by detecting this scenario, hiding the duplicates, and logging an error, similar to how this is handled in `rclone sync`. ## VFS Disk Options This flag allows you to manually set the statistics about the filing system. It can be useful when those statistics cannot be read correctly automatically. --vfs-disk-space-total-size Manually set the total disk space size (example: 256G, default: -1) ## Alternate report of used bytes Some backends, most notably S3, do not report the amount of bytes used. If you need this information to be available when running `df` on the filesystem, then pass the flag `--vfs-used-is-size` to rclone. With this flag set, instead of relying on the backend to report this information, rclone will scan the whole remote similar to `rclone size` and compute the total used space itself. _WARNING._ Contrary to `rclone size`, this flag ignores filters so that the result is accurate. However, this is very inefficient and may cost lots of API calls resulting in extra charges. Use it as a last resort and only with caching. ``` rclone serve s3 remote:path [flags] ``` ## Options ``` --addr stringArray IPaddress:Port or :Port to bind server to (default [127.0.0.1:8080]) --allow-origin string Origin which cross-domain request (CORS) can be executed from --auth-key stringArray Set key pair for v4 authorization: access_key_id,secret_access_key --baseurl string Prefix for URLs - leave blank for root --cert string TLS PEM key (concatenation of certificate and CA certificate) --client-ca string Client certificate authority to verify clients with --dir-cache-time Duration Time to cache directory entries for (default 5m0s) --dir-perms FileMode Directory permissions (default 0777) --etag-hash string Which hash to use for the ETag, or auto or blank for off (default "MD5") --file-perms FileMode File permissions (default 0666) --force-path-style If true use path style access if false use virtual hosted style (default true) (default true) --gid uint32 Override the gid field set by the filesystem (not supported on Windows) (default 1000) -h, --help help for s3 --key string TLS PEM Private key --max-header-bytes int Maximum size of request header (default 4096) --min-tls-version string Minimum TLS version that is acceptable (default "tls1.0") --no-checksum Don't compare checksums on up/download --no-cleanup Not to cleanup empty folder after object is deleted --no-modtime Don't read/write the modification time (can speed things up) --no-seek Don't allow seeking in files --poll-interval Duration Time to wait between polling for changes, must be smaller than dir-cache-time and only on supported remotes (set 0 to disable) (default 1m0s) --read-only Only allow read-only access --server-read-timeout Duration Timeout for server reading data (default 1h0m0s) --server-write-timeout Duration Timeout for server writing data (default 1h0m0s) --uid uint32 Override the uid field set by the filesystem (not supported on Windows) (default 1000) --umask int Override the permission bits set by the filesystem (not supported on Windows) (default 2) --vfs-block-norm-dupes If duplicate filenames exist in the same directory (after normalization), log an error and hide the duplicates (may have a performance cost) --vfs-cache-max-age Duration Max time since last access of objects in the cache (default 1h0m0s) --vfs-cache-max-size SizeSuffix Max total size of objects in the cache (default off) --vfs-cache-min-free-space SizeSuffix Target minimum free space on the disk containing the cache (default off) --vfs-cache-mode CacheMode Cache mode off|minimal|writes|full (default off) --vfs-cache-poll-interval Duration Interval to poll the cache for stale objects (default 1m0s) --vfs-case-insensitive If a file name not found, find a case insensitive match --vfs-disk-space-total-size SizeSuffix Specify the total space of disk (default off) --vfs-fast-fingerprint Use fast (less accurate) fingerprints for change detection --vfs-read-ahead SizeSuffix Extra read ahead over --buffer-size when using cache-mode full --vfs-read-chunk-size SizeSuffix Read the source objects in chunks (default 128Mi) --vfs-read-chunk-size-limit SizeSuffix If greater than --vfs-read-chunk-size, double the chunk size after each chunk read, until the limit is reached ('off' is unlimited) (default off) --vfs-read-wait Duration Time to wait for in-sequence read before seeking (default 20ms) --vfs-refresh Refreshes the directory cache recursively in the background on start --vfs-used-is-size rclone size Use the rclone size algorithm for Used size --vfs-write-back Duration Time to writeback files after last use when using cache (default 5s) --vfs-write-wait Duration Time to wait for in-sequence write before giving error (default 1s) ``` ## Filter Options Flags for filtering directory listings. ``` --delete-excluded Delete files on dest excluded from sync --exclude stringArray Exclude files matching pattern --exclude-from stringArray Read file exclude patterns from file (use - to read from stdin) --exclude-if-present stringArray Exclude directories if filename is present --files-from stringArray Read list of source-file names from file (use - to read from stdin) --files-from-raw stringArray Read list of source-file names from file without any processing of lines (use - to read from stdin) -f, --filter stringArray Add a file filtering rule --filter-from stringArray Read file filtering patterns from a file (use - to read from stdin) --ignore-case Ignore case in filters (case insensitive) --include stringArray Include files matching pattern --include-from stringArray Read file include patterns from file (use - to read from stdin) --max-age Duration Only transfer files younger than this in s or suffix ms|s|m|h|d|w|M|y (default off) --max-depth int If set limits the recursion depth to this (default -1) --max-size SizeSuffix Only transfer files smaller than this in KiB or suffix B|K|M|G|T|P (default off) --metadata-exclude stringArray Exclude metadatas matching pattern --metadata-exclude-from stringArray Read metadata exclude patterns from file (use - to read from stdin) --metadata-filter stringArray Add a metadata filtering rule --metadata-filter-from stringArray Read metadata filtering patterns from a file (use - to read from stdin) --metadata-include stringArray Include metadatas matching pattern --metadata-include-from stringArray Read metadata include patterns from file (use - to read from stdin) --min-age Duration Only transfer files older than this in s or suffix ms|s|m|h|d|w|M|y (default off) --min-size SizeSuffix Only transfer files bigger than this in KiB or suffix B|K|M|G|T|P (default off) ``` See the [global flags page](/flags/) for global options not listed here. # SEE ALSO * [rclone serve](/commands/rclone_serve/) - Serve a remote over a protocol.