add support for conditional resuming of uploads

Signed-off-by: Nicola Murino <[email protected]>

docs/full-configuration.md

@@ -62,13 +62,14 @@ The configuration file contains the following sections:
 
 - **"common"**, configuration parameters shared among all the supported protocols
   - `idle_timeout`, integer. Time in minutes after which an idle client will be disconnected. 0 means disabled. Default: 15
-  - `upload_mode` integer. 0 means standard: the files are uploaded directly to the requested path. 1 means atomic: files are uploaded to a temporary path and renamed to the requested path when the client ends the upload. Atomic mode avoids problems such as a web server that serves partial files when the files are being uploaded. In atomic mode, if there is an upload error, the temporary file is deleted and so the requested upload path will not contain a partial file. 2 means atomic with resume support: same as atomic but if there is an upload error, the temporary file is renamed to the requested path and not deleted. This way, a client can reconnect and resume the upload. Ignored for cloud-based storage backends (uploads are always atomic and resume is not supported for these backends) and for SFTP backend if buffering is enabled. Default: 0
+  - `upload_mode` integer. 0 means standard: the files are uploaded directly to the requested path. 1 means atomic: files are uploaded to a temporary path and renamed to the requested path when the client ends the upload. Atomic mode avoids problems such as a web server that serves partial files when the files are being uploaded. In atomic mode, if there is an upload error, the temporary file is deleted and so the requested upload path will not contain a partial file. 2 means atomic with resume support: same as atomic but if there is an upload error, the temporary file is renamed to the requested path and not deleted. This way, a client can reconnect and resume the upload. Ignored for cloud-based storage backends (uploads are always atomic and upload resume is not supported, by default, for these backends) and for SFTP backend if buffering is enabled. Default: `0`
   - `actions`, struct. It contains the command to execute and/or the HTTP URL to notify and the trigger conditions. See [Custom Actions](./custom-actions.md) for more details
     - `execute_on`, list of strings. Valid values are `pre-download`, `download`, `first-download`, `pre-upload`, `upload`, `first-upload`, `pre-delete`, `delete`, `rename`, `mkdir`, `rmdir`, `ssh_cmd`, `copy`. Leave empty to disable actions.
     - `execute_sync`, list of strings. Actions, defined in the `execute_on` list above, to be performed synchronously. The `pre-*` actions are always executed synchronously while the other ones are asynchronous. Executing an action synchronously means that SFTPGo will not return a result code to the client (which is waiting for it) until your hook have completed its execution. Leave empty to execute only the defined `pre-*` hook synchronously
     - `hook`, string. Absolute path to the command to execute or HTTP URL to notify.
   - `setstat_mode`, integer. 0 means "normal mode": requests for changing permissions, owner/group and access/modification times are executed. 1 means "ignore mode": requests for changing permissions, owner/group and access/modification times are silently ignored. 2 means "ignore mode if not supported": requests for changing permissions and owner/group are silently ignored for cloud filesystems and executed for local/SFTP filesystem. Requests for changing modification times are always executed for local/SFTP filesystems and are executed for cloud based filesystems if the target is a file and there is a metadata plugin available. A metadata plugin can be found [here](https://github.com/sftpgo/sftpgo-plugin-metadata).
   - `rename_mode`, integer. By default (`0`), renaming of non-empty directories is not allowed for cloud storage providers (S3, GCS, Azure Blob). Set to `1` to enable recursive renames for these providers, they may be slow, there is no atomic rename API like for local filesystem, so SFTPGo will recursively list the directory contents and do a rename for each entry (partial renaming and incorrect disk quota updates are possible in error cases). Default `0`.
+  - `resume_max_size`, integer. defines the maximum size allowed, in bytes, to resume uploads on storage backends with immutable objects. By default, resuming uploads is not allowed for cloud storage providers (S3, GCS, Azure Blob) because SFTPGo must rewrite the entire file. Set to a value greater than 0 to allow resuming uploads of files smaller than or equal to the defined size. Please note that uploads for these backends are still atomic, the client must intentionally upload a portion of the target file and then resume uploading.. Default `0`.
   - `temp_path`, string. Defines the path for temporary files such as those used for atomic uploads or file pipes. If you set this option you must make sure that the defined path exists, is accessible for writing by the user running SFTPGo, and is on the same filesystem as the users home directories otherwise the renaming for atomic uploads will become a copy and therefore may take a long time. The temporary files are not namespaced. The default is generally fine. Leave empty for the default.
   - `proxy_protocol`, integer. Support for [HAProxy PROXY protocol](https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt). If you are running SFTPGo behind a proxy server such as HAProxy, AWS ELB or NGINX, you can enable the proxy protocol. It provides a convenient way to safely transport connection information such as a client's address across multiple layers of NAT or TCP proxies to get the real client IP address instead of the proxy IP. Both protocol versions 1 and 2 are supported. If the proxy protocol is enabled in SFTPGo then you have to enable the protocol in your proxy configuration too. For example, for HAProxy, add `send-proxy` or `send-proxy-v2` to each server configuration line. The PROXY protocol is supported for SSH/SFTP and FTP/S. The following modes are supported:
     - 0, disabled

go.mod

@@ -7,7 +7,7 @@ require (
 	github.com/Azure/azure-sdk-for-go/sdk/azcore v1.8.0
 	github.com/Azure/azure-sdk-for-go/sdk/storage/azblob v1.2.0
 	github.com/GehirnInc/crypt v0.0.0-20230320061759-8cc1b52080c5
-	github.com/alexedwards/argon2id v0.0.0-20231016161201-27bf9713919b
+	github.com/alexedwards/argon2id v1.0.0
 	github.com/amoghe/go-crypt v0.0.0-20220222110647-20eada5f5964
 	github.com/aws/aws-sdk-go-v2 v1.21.2
 	github.com/aws/aws-sdk-go-v2/config v1.19.0
@@ -18,7 +18,7 @@ require (
 	github.com/aws/aws-sdk-go-v2/service/s3 v1.40.2
 	github.com/aws/aws-sdk-go-v2/service/secretsmanager v1.21.6
 	github.com/aws/aws-sdk-go-v2/service/sts v1.23.2
-	github.com/bmatcuk/doublestar/v4 v4.6.0
+	github.com/bmatcuk/doublestar/v4 v4.6.1
 	github.com/cockroachdb/cockroach-go/v2 v2.3.5
 	github.com/coreos/go-oidc/v3 v3.7.0
 	github.com/drakkan/webdav v0.0.0-20230227175313-32996838bcd8
@@ -106,7 +106,7 @@ require (
 	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect
 	github.com/decred/dcrd/dcrec/secp256k1/v4 v4.2.0 // indirect
 	github.com/fatih/color v1.15.0 // indirect
-	github.com/fsnotify/fsnotify v1.6.0 // indirect
+	github.com/fsnotify/fsnotify v1.7.0 // indirect
 	github.com/go-jose/go-jose/v3 v3.0.0 // indirect
 	github.com/go-ole/go-ole v1.3.0 // indirect
 	github.com/goccy/go-json v0.10.2 // indirect

go.sum

@@ -67,8 +67,8 @@ github.com/GehirnInc/crypt v0.0.0-20230320061759-8cc1b52080c5 h1:IEjq88XO4PuBDcv
 github.com/GehirnInc/crypt v0.0.0-20230320061759-8cc1b52080c5/go.mod h1:exZ0C/1emQJAw5tHOaUDyY1ycttqBAPcxuzf7QbY6ec=
 github.com/ajg/form v1.5.1 h1:t9c7v8JUKu/XxOGBU0yjNpaMloxGEJhUkqFRq0ibGeU=
 github.com/ajg/form v1.5.1/go.mod h1:uL1WgH+h2mgNtvBq0339dVnzXdBETtL2LeUXaIv25UY=
-github.com/alexedwards/argon2id v0.0.0-20231016161201-27bf9713919b h1:11BUm+H/afKXfPQ7byMNdYxHlXmBP9g1Smo72gz3c4o=
-github.com/alexedwards/argon2id v0.0.0-20231016161201-27bf9713919b/go.mod h1:mTeFRcTdnpzOlRjMoFYC/80HwVUreupyAiqPkCZQOXc=
+github.com/alexedwards/argon2id v1.0.0 h1:wJzDx66hqWX7siL/SRUmgz3F8YMrd/nfX/xHHcQQP0w=
+github.com/alexedwards/argon2id v1.0.0/go.mod h1:tYKkqIjzXvZdzPvADMWOEZ+l6+BD6CtBXMj5fnJppiw=
 github.com/amoghe/go-crypt v0.0.0-20220222110647-20eada5f5964 h1:I9YN9WMo3SUh7p/4wKeNvD/IQla3U3SUa61U7ul+xM4=
 github.com/amoghe/go-crypt v0.0.0-20220222110647-20eada5f5964/go.mod h1:eFiR01PwTcpbzXtdMces7zxg6utvFM5puiWHpWB8D/k=
 github.com/aws/aws-sdk-go-v2 v1.21.2 h1:+LXZ0sgo8quN9UOKXXzAWRT3FWd4NxeXWOZom9pE7GA=
@@ -115,8 +115,8 @@ github.com/aws/smithy-go v1.15.0 h1:PS/durmlzvAFpQHDs4wi4sNNP9ExsqZh6IlfdHXgKK8=
 github.com/aws/smithy-go v1.15.0/go.mod h1:Tg+OJXh4MB2R/uN61Ko2f6hTZwB/ZYGOtib8J3gBHzA=
 github.com/beorn7/perks v1.0.1 h1:VlbKKnNfV8bJzeqoa4cOKqO6bYr3WgKZxO8Z16+hsOM=
 github.com/beorn7/perks v1.0.1/go.mod h1:G2ZrVWU2WbWT9wwq4/hrbKbnv/1ERSJQ0ibhJ6rlkpw=
-github.com/bmatcuk/doublestar/v4 v4.6.0 h1:HTuxyug8GyFbRkrffIpzNCSK4luc0TY3wzXvzIZhEXc=
-github.com/bmatcuk/doublestar/v4 v4.6.0/go.mod h1:xBQ8jztBU6kakFMg+8WGxn0c6z1fTSPVIjEY1Wr7jzc=
+github.com/bmatcuk/doublestar/v4 v4.6.1 h1:FH9SifrbvJhnlQpztAx++wlkk70QBf0iBWDwNy7PA4I=
+github.com/bmatcuk/doublestar/v4 v4.6.1/go.mod h1:xBQ8jztBU6kakFMg+8WGxn0c6z1fTSPVIjEY1Wr7jzc=
 github.com/boombuler/barcode v1.0.1-0.20190219062509-6c824513bacc/go.mod h1:paBWMcWSl3LHKBqUq+rly7CNSldXjb2rDl3JlRe0mD8=
 github.com/boombuler/barcode v1.0.1 h1:NDBbPmhS+EqABEs5Kg3n/5ZNjy73Pz7SIV+KCeqyXcs=
 github.com/boombuler/barcode v1.0.1/go.mod h1:paBWMcWSl3LHKBqUq+rly7CNSldXjb2rDl3JlRe0mD8=
@@ -177,8 +177,8 @@ github.com/fclairamb/go-log v0.4.1 h1:rLtdSG9x2pK41AIAnE8WYpl05xBJfw1ZyYxZaXFcBs
 github.com/fclairamb/go-log v0.4.1/go.mod h1:sw1KvnkZ4wKCYkvy4SL3qVZcJSWFP8Ure4pM3z+KNn4=
 github.com/frankban/quicktest v1.14.4 h1:g2rn0vABPOOXmZUj+vbmUp0lPoXEMuhTpIluN0XL9UY=
 github.com/frankban/quicktest v1.14.4/go.mod h1:4ptaffx2x8+WTWXmUCuVU6aPUX1/Mz7zb5vbUoiM6w0=
-github.com/fsnotify/fsnotify v1.6.0 h1:n+5WquG0fcWoWp6xPWfHdbskMCQaFnG6PfBrh1Ky4HY=
-github.com/fsnotify/fsnotify v1.6.0/go.mod h1:sl3t1tCWJFWoRz9R8WJCbQihKKwmorjAbSClcnxKAGw=
+github.com/fsnotify/fsnotify v1.7.0 h1:8JEhPFa5W2WU7YfeZzPNqzMP6Lwt7L2715Ggo0nosvA=
+github.com/fsnotify/fsnotify v1.7.0/go.mod h1:40Bi/Hjc2AVfZrqy+aj+yEI+/bRxZnMJyTJwOpGvigM=
 github.com/go-acme/lego/v4 v4.14.2 h1:/D/jqRgLi8Cbk33sLGtu2pX2jEg3bGJWHyV8kFuUHGM=
 github.com/go-acme/lego/v4 v4.14.2/go.mod h1:kBXxbeTg0x9AgaOYjPSwIeJy3Y33zTz+tMD16O4MO6c=
 github.com/go-chi/chi/v5 v5.0.10 h1:rLz5avzKpjqxrYwXNfmjkrYYXOyLJd37pz53UFHC6vk=
@@ -670,7 +670,6 @@ golang.org/x/sys v0.0.0-20220520151302-bc2c85ada10a/go.mod h1:oPkhp1MJrh7nUepCBc
 golang.org/x/sys v0.0.0-20220704084225-05e143d24a9e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220722155257-8c9f86f7a55f/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.0.0-20220811171246-fbc7d0a398ab/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.0.0-20220908164124-27713097b956/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.1.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.5.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=

internal/common/common.go

@@ -230,6 +230,7 @@ func Initialize(c Configuration, isShared int) error {
 	vfs.SetAllowSelfConnections(c.AllowSelfConnections)
 	vfs.SetRenameMode(c.RenameMode)
 	vfs.SetReadMetadataMode(c.Metadata.Read)
+	vfs.SetResumeMaxSize(c.ResumeMaxSize)
 	dataprovider.SetAllowSelfConnections(c.AllowSelfConnections)
 	transfersChecker = getTransfersChecker(isShared)
 	return nil
@@ -523,6 +524,12 @@ type Configuration struct {
 	// renames for these providers, they may be slow, there is no atomic rename API like for local
 	// filesystem, so SFTPGo will recursively list the directory contents and do a rename for each entry
 	RenameMode int `json:"rename_mode" mapstructure:"rename_mode"`
+	// ResumeMaxSize defines the maximum size allowed, in bytes, to resume uploads on storage backends
+	// with immutable objects. By default, resuming uploads is not allowed for cloud storage providers
+	// (S3, GCS, Azure Blob) because SFTPGo must rewrite the entire file.
+	// Set to a value greater than 0 to allow resuming uploads of files smaller than or equal to the
+	// defined size.
+	ResumeMaxSize int64 `json:"resume_max_size" mapstructure:"resume_max_size"`
 	// TempPath defines the path for temporary files such as those used for atomic uploads or file pipes.
 	// If you set this option you must make sure that the defined path exists, is accessible for writing
 	// by the user running SFTPGo, and is on the same filesystem as the users home directories otherwise

internal/common/connection.go

@@ -343,14 +343,19 @@ func (c *BaseConnection) CheckParentDirs(virtualPath string) error {
 }
 
 // GetCreateChecks returns the checks for creating new files
-func (c *BaseConnection) GetCreateChecks(virtualPath string, isNewFile bool) int {
+func (c *BaseConnection) GetCreateChecks(virtualPath string, isNewFile bool, isResume bool) int {
+	result := 0
 	if !isNewFile {
-		return 0
+		if isResume {
+			result += vfs.CheckResume
+		}
+		return result
 	}
 	if !c.User.HasPerm(dataprovider.PermCreateDirs, path.Dir(virtualPath)) {
-		return vfs.CheckParentDir
+		result += vfs.CheckParentDir
+		return result
 	}
-	return 0
+	return result
 }
 
 // CreateDir creates a new directory at the specified fsPath

internal/common/eventmanager.go

@@ -955,7 +955,7 @@ func getFileWriter(conn *BaseConnection, virtualPath string, expectedSize int64)
 	if err := checkWriterPermsAndQuota(conn, virtualPath, numFiles, expectedSize, truncatedSize); err != nil {
 		return nil, numFiles, truncatedSize, nil, err
 	}
-	f, w, cancelFn, err := fs.Create(fsPath, 0, conn.GetCreateChecks(virtualPath, numFiles == 1))
+	f, w, cancelFn, err := fs.Create(fsPath, 0, conn.GetCreateChecks(virtualPath, numFiles == 1, false))
 	if err != nil {
 		return nil, numFiles, truncatedSize, nil, conn.GetFsError(fs, err)
 	}

internal/config/config.go

@@ -203,6 +203,7 @@ func Init() {
 			},
 			SetstatMode:           0,
 			RenameMode:            0,
+			ResumeMaxSize:         0,
 			TempPath:              "",
 			ProxyProtocol:         0,
 			ProxyAllowed:          []string{},
@@ -1993,6 +1994,7 @@ func setViperDefaults() {
 	viper.SetDefault("common.actions.hook", globalConf.Common.Actions.Hook)
 	viper.SetDefault("common.setstat_mode", globalConf.Common.SetstatMode)
 	viper.SetDefault("common.rename_mode", globalConf.Common.RenameMode)
+	viper.SetDefault("common.resume_max_size", globalConf.Common.ResumeMaxSize)
 	viper.SetDefault("common.temp_path", globalConf.Common.TempPath)
 	viper.SetDefault("common.proxy_protocol", globalConf.Common.ProxyProtocol)
 	viper.SetDefault("common.proxy_allowed", globalConf.Common.ProxyAllowed)

internal/ftpd/handler.go

@@ -408,7 +408,7 @@ func (c *Connection) handleFTPUploadToNewFile(fs vfs.Fs, flags int, resolvedPath
 		c.Log(logger.LevelDebug, "upload for file %q denied by pre action: %v", requestPath, err)
 		return nil, ftpserver.ErrFileNameNotAllowed
 	}
-	file, w, cancelFn, err := fs.Create(filePath, flags, c.GetCreateChecks(requestPath, true))
+	file, w, cancelFn, err := fs.Create(filePath, flags, c.GetCreateChecks(requestPath, true, false))
 	if err != nil {
 		c.Log(logger.LevelError, "error creating file %q, flags %v: %+v", resolvedPath, flags, err)
 		return nil, c.GetFsError(fs, err)
@@ -444,7 +444,7 @@ func (c *Connection) handleFTPUploadToExistingFile(fs vfs.Fs, flags int, resolve
 	isResume := flags&os.O_TRUNC == 0
 	// if there is a size limit remaining size cannot be 0 here, since quotaResult.HasSpace
 	// will return false in this case and we deny the upload before
-	maxWriteSize, err := c.GetMaxWriteSize(diskQuota, isResume, fileSize, fs.IsUploadResumeSupported())
+	maxWriteSize, err := c.GetMaxWriteSize(diskQuota, isResume, fileSize, vfs.IsUploadResumeSupported(fs, fileSize))
 	if err != nil {
 		c.Log(logger.LevelDebug, "unable to get max write size: %v", err)
 		return nil, err
@@ -463,7 +463,7 @@ func (c *Connection) handleFTPUploadToExistingFile(fs vfs.Fs, flags int, resolve
 		}
 	}
 
-	file, w, cancelFn, err := fs.Create(filePath, flags, c.GetCreateChecks(requestPath, false))
+	file, w, cancelFn, err := fs.Create(filePath, flags, c.GetCreateChecks(requestPath, false, isResume))
 	if err != nil {
 		c.Log(logger.LevelError, "error opening existing file, flags: %v, source: %q, err: %+v", flags, filePath, err)
 		return nil, c.GetFsError(fs, err)

internal/ftpd/internal_test.go

@@ -367,6 +367,12 @@ func (MockOsFs) IsUploadResumeSupported() bool {
 	return false
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (MockOsFs) IsConditionalUploadResumeSupported(_ int64) bool {
+	return false
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported
 func (fs MockOsFs) IsAtomicUploadSupported() bool {
 	return fs.isAtomicUploadSupported

internal/httpd/handler.go

@@ -201,7 +201,7 @@ func (c *Connection) handleUploadFile(fs vfs.Fs, resolvedPath, filePath, request
 
 	maxWriteSize, _ := c.GetMaxWriteSize(diskQuota, false, fileSize, fs.IsUploadResumeSupported())
 
-	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.GetCreateChecks(requestPath, isNewFile))
+	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.GetCreateChecks(requestPath, isNewFile, false))
 	if err != nil {
 		c.Log(logger.LevelError, "error opening existing file, source: %q, err: %+v", filePath, err)
 		return nil, c.GetFsError(fs, err)

internal/sftpd/handler.go

@@ -407,7 +407,7 @@ func (c *Connection) handleSFTPUploadToNewFile(fs vfs.Fs, pflags sftp.FileOpenFl
 	}
 
 	osFlags := getOSOpenFlags(pflags)
-	file, w, cancelFn, err := fs.Create(filePath, osFlags, c.GetCreateChecks(requestPath, true))
+	file, w, cancelFn, err := fs.Create(filePath, osFlags, c.GetCreateChecks(requestPath, true, false))
 	if err != nil {
 		c.Log(logger.LevelError, "error creating file %q, os flags %d, pflags %+v: %+v", resolvedPath, osFlags, pflags, err)
 		return nil, c.GetFsError(fs, err)
@@ -443,7 +443,7 @@ func (c *Connection) handleSFTPUploadToExistingFile(fs vfs.Fs, pflags sftp.FileO
 	// if there is a size limit the remaining size cannot be 0 here, since quotaResult.HasSpace
 	// will return false in this case and we deny the upload before.
 	// For Cloud FS GetMaxWriteSize will return unsupported operation
-	maxWriteSize, err := c.GetMaxWriteSize(diskQuota, isResume, fileSize, fs.IsUploadResumeSupported())
+	maxWriteSize, err := c.GetMaxWriteSize(diskQuota, isResume, fileSize, vfs.IsUploadResumeSupported(fs, fileSize))
 	if err != nil {
 		c.Log(logger.LevelDebug, "unable to get max write size for file %q is resume? %t: %v",
 			requestPath, isResume, err)
@@ -464,7 +464,7 @@ func (c *Connection) handleSFTPUploadToExistingFile(fs vfs.Fs, pflags sftp.FileO
 		}
 	}
 
-	file, w, cancelFn, err := fs.Create(filePath, osFlags, c.GetCreateChecks(requestPath, false))
+	file, w, cancelFn, err := fs.Create(filePath, osFlags, c.GetCreateChecks(requestPath, false, isResume))
 	if err != nil {
 		c.Log(logger.LevelError, "error opening existing file, os flags %v, pflags: %+v, source: %q, err: %+v",
 			osFlags, pflags, filePath, err)
@@ -476,22 +476,15 @@ func (c *Connection) handleSFTPUploadToExistingFile(fs vfs.Fs, pflags sftp.FileO
 	if isResume {
 		c.Log(logger.LevelDebug, "resuming upload requested, file path %q initial size: %d, has append flag %t",
 			filePath, fileSize, pflags.Append)
-		// enforce min write offset only if the client passed the APPEND flag
-		if pflags.Append {
+		// enforce min write offset only if the client passed the APPEND flag or the filesystem
+		// supports emulated resume
+		if pflags.Append || !fs.IsUploadResumeSupported() {
 			minWriteOffset = fileSize
 		}
 		initialSize = fileSize
 	} else {
 		if isTruncate && vfs.HasTruncateSupport(fs) {
-			vfolder, err := c.User.GetVirtualFolderForPath(path.Dir(requestPath))
-			if err == nil {
-				dataprovider.UpdateVirtualFolderQuota(&vfolder.BaseVirtualFolder, 0, -fileSize, false) //nolint:errcheck
-				if vfolder.IsIncludedInUserQuota() {
-					dataprovider.UpdateUserQuota(&c.User, 0, -fileSize, false) //nolint:errcheck
-				}
-			} else {
-				dataprovider.UpdateUserQuota(&c.User, 0, -fileSize, false) //nolint:errcheck
-			}
+			c.updateQuotaAfterTruncate(requestPath, fileSize)
 		} else {
 			initialSize = fileSize
 			truncatedSize = fileSize
@@ -562,6 +555,18 @@ func (c *Connection) getStatVFSFromQuotaResult(fs vfs.Fs, name string, quotaResu
 	}, nil
 }
 
+func (c *Connection) updateQuotaAfterTruncate(requestPath string, fileSize int64) {
+	vfolder, err := c.User.GetVirtualFolderForPath(path.Dir(requestPath))
+	if err == nil {
+		dataprovider.UpdateVirtualFolderQuota(&vfolder.BaseVirtualFolder, 0, -fileSize, false) //nolint:errcheck
+		if vfolder.IsIncludedInUserQuota() {
+			dataprovider.UpdateUserQuota(&c.User, 0, -fileSize, false) //nolint:errcheck
+		}
+	} else {
+		dataprovider.UpdateUserQuota(&c.User, 0, -fileSize, false) //nolint:errcheck
+	}
+}
+
 func getOSOpenFlags(requestFlags sftp.FileOpenFlags) (flags int) {
 	var osFlags int
 	if requestFlags.Read && requestFlags.Write {

internal/sftpd/internal_test.go

@@ -109,6 +109,12 @@ func (MockOsFs) IsUploadResumeSupported() bool {
 	return false
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (MockOsFs) IsConditionalUploadResumeSupported(_ int64) bool {
+	return false
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported
 func (fs MockOsFs) IsAtomicUploadSupported() bool {
 	return fs.isAtomicUploadSupported

internal/sftpd/scp.go

@@ -245,7 +245,7 @@ func (c *scpCommand) handleUploadFile(fs vfs.Fs, resolvedPath, filePath string,
 
 	maxWriteSize, _ := c.connection.GetMaxWriteSize(diskQuota, false, fileSize, fs.IsUploadResumeSupported())
 
-	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.connection.GetCreateChecks(requestPath, isNewFile))
+	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.connection.GetCreateChecks(requestPath, isNewFile, false))
 	if err != nil {
 		c.connection.Log(logger.LevelError, "error creating file %q: %v", resolvedPath, err)
 		c.sendErrorMessage(fs, err)

internal/vfs/azblobfs.go

@@ -264,6 +264,23 @@ func (fs *AzureBlobFs) Create(name string, flag, checks int) (File, *PipeWriter,
 		metric.AZTransferCompleted(r.GetReadedBytes(), 0, err)
 	}()
 
+	if checks&CheckResume != 0 {
+		readCh := make(chan error, 1)
+
+		go func() {
+			err = fs.downloadToWriter(name, p)
+			readCh <- err
+		}()
+
+		err = <-readCh
+		if err != nil {
+			cancelFn()
+			p.Close()
+			fsLog(fs, logger.LevelDebug, "download before resume failed, writer closed and read cancelled")
+			return nil, nil, nil, err
+		}
+	}
+
 	return nil, p, cancelFn, nil
 }
 
@@ -458,6 +475,12 @@ func (*AzureBlobFs) IsUploadResumeSupported() bool {
 	return false
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (*AzureBlobFs) IsConditionalUploadResumeSupported(size int64) bool {
+	return size <= resumeMaxSize
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported.
 // Azure Blob uploads are already atomic, we don't need to upload to a temporary
 // file
@@ -965,7 +988,7 @@ func (fs *AzureBlobFs) handleMultipartDownload(ctx context.Context, blockBlob *b
 		fsLog(fs, logger.LevelError, "unable to get blob properties, download aborted: %+v", err)
 		return err
 	}
-	if readMetadata > 0 {
+	if readMetadata > 0 && pipeReader != nil {
 		pipeReader.setMetadataFromPointerVal(props.Metadata)
 	}
 	contentLength := util.GetIntFromPointer(props.ContentLength)
@@ -1172,6 +1195,19 @@ func (fs *AzureBlobFs) getCopyOptions() *blob.StartCopyFromURLOptions {
 	return copyOptions
 }
 
+func (fs *AzureBlobFs) downloadToWriter(name string, w *PipeWriter) error {
+	fsLog(fs, logger.LevelDebug, "starting download before resuming upload, path %q", name)
+	ctx, cancelFn := context.WithTimeout(context.Background(), preResumeTimeout)
+	defer cancelFn()
+
+	blockBlob := fs.containerClient.NewBlockBlobClient(name)
+	err := fs.handleMultipartDownload(ctx, blockBlob, 0, w, nil)
+	fsLog(fs, logger.LevelDebug, "download before resuming upload completed, path %q size: %d, err: %+v",
+		name, w.GetWrittenBytes(), err)
+	metric.AZTransferCompleted(w.GetWrittenBytes(), 1, err)
+	return err
+}
+
 func (fs *AzureBlobFs) getStorageID() string {
 	if fs.config.Endpoint != "" {
 		if !strings.HasSuffix(fs.config.Endpoint, "/") {

internal/vfs/cryptfs.go

@@ -243,6 +243,12 @@ func (*CryptFs) IsUploadResumeSupported() bool {
 	return false
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (*CryptFs) IsConditionalUploadResumeSupported(_ int64) bool {
+	return false
+}
+
 // GetMimeType returns the content type
 func (fs *CryptFs) GetMimeType(name string) (string, error) {
 	f, key, err := fs.getFileAndEncryptionKey(name)

internal/vfs/gcsfs.go

@@ -178,7 +178,6 @@ func (fs *GCSFs) Create(name string, flag, checks int) (File, *PipeWriter, func(
 	if err != nil {
 		return nil, nil, nil, err
 	}
-	p := NewPipeWriter(w)
 	bkt := fs.svc.Bucket(fs.config.Bucket)
 	obj := bkt.Object(name)
 	if flag == -1 {
@@ -193,6 +192,7 @@ func (fs *GCSFs) Create(name string, flag, checks int) (File, *PipeWriter, func(
 			fsLog(fs, logger.LevelWarn, "unable to set precondition for %q, stat err: %v", name, statErr)
 		}
 	}
+	p := NewPipeWriter(w)
 
 	ctx, cancelFn := context.WithCancel(context.Background())
 	objectWriter := obj.NewWriter(ctx)
@@ -208,15 +208,8 @@ func (fs *GCSFs) Create(name string, flag, checks int) (File, *PipeWriter, func(
 	} else {
 		contentType = mime.TypeByExtension(path.Ext(name))
 	}
-	if contentType != "" {
-		objectWriter.ObjectAttrs.ContentType = contentType
-	}
-	if fs.config.StorageClass != "" {
-		objectWriter.ObjectAttrs.StorageClass = fs.config.StorageClass
-	}
-	if fs.config.ACL != "" {
-		objectWriter.PredefinedACL = fs.config.ACL
-	}
+	fs.setWriterAttrs(objectWriter, contentType)
+
 	go func() {
 		defer cancelFn()
 
@@ -231,6 +224,24 @@ func (fs *GCSFs) Create(name string, flag, checks int) (File, *PipeWriter, func(
 			name, fs.config.ACL, n, err)
 		metric.GCSTransferCompleted(n, 0, err)
 	}()
+
+	if checks&CheckResume != 0 {
+		readCh := make(chan error, 1)
+
+		go func() {
+			err = fs.downloadToWriter(name, p)
+			readCh <- err
+		}()
+
+		err = <-readCh
+		if err != nil {
+			cancelFn()
+			p.Close()
+			fsLog(fs, logger.LevelDebug, "download before resume failed, writer closed and read cancelled")
+			return nil, nil, nil, err
+		}
+	}
+
 	return nil, p, cancelFn, nil
 }
 
@@ -429,6 +440,12 @@ func (*GCSFs) IsUploadResumeSupported() bool {
 	return false
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (*GCSFs) IsConditionalUploadResumeSupported(size int64) bool {
+	return size <= resumeMaxSize
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported.
 // S3 uploads are already atomic, we don't need to upload to a temporary
 // file
@@ -748,6 +765,37 @@ func (fs *GCSFs) getObjectStat(name string) (os.FileInfo, error) {
 	return updateFileInfoModTime(fs.getStorageID(), name, NewFileInfo(name, true, attrs.Size, attrs.Updated, false))
 }
 
+func (fs *GCSFs) setWriterAttrs(objectWriter *storage.Writer, contentType string) {
+	if contentType != "" {
+		objectWriter.ObjectAttrs.ContentType = contentType
+	}
+	if fs.config.StorageClass != "" {
+		objectWriter.ObjectAttrs.StorageClass = fs.config.StorageClass
+	}
+	if fs.config.ACL != "" {
+		objectWriter.PredefinedACL = fs.config.ACL
+	}
+}
+
+func (fs *GCSFs) downloadToWriter(name string, w *PipeWriter) error {
+	fsLog(fs, logger.LevelDebug, "starting download before resuming upload, path %q", name)
+	ctx, cancelFn := context.WithTimeout(context.Background(), preResumeTimeout)
+	defer cancelFn()
+
+	bkt := fs.svc.Bucket(fs.config.Bucket)
+	obj := bkt.Object(name)
+	objectReader, err := obj.NewRangeReader(ctx, 0, -1)
+	if err != nil {
+		fsLog(fs, logger.LevelDebug, "unable to start download before resuming upload, path %q, err: %v", name, err)
+		return err
+	}
+	n, err := io.Copy(w, objectReader)
+	fsLog(fs, logger.LevelDebug, "download before resuming upload completed, path %q size: %d, err: %+v",
+		name, n, err)
+	metric.GCSTransferCompleted(n, 1, err)
+	return err
+}
+
 func (fs *GCSFs) copyFileInternal(source, target string) error {
 	src := fs.svc.Bucket(fs.config.Bucket).Object(source)
 	dst := fs.svc.Bucket(fs.config.Bucket).Object(target)

internal/vfs/httpfs.go

@@ -501,6 +501,12 @@ func (*HTTPFs) IsUploadResumeSupported() bool {
 	return false
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (*HTTPFs) IsConditionalUploadResumeSupported(_ int64) bool {
+	return false
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported.
 func (*HTTPFs) IsAtomicUploadSupported() bool {
 	return false

internal/vfs/osfs.go

@@ -279,6 +279,12 @@ func (*OsFs) IsUploadResumeSupported() bool {
 	return true
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (*OsFs) IsConditionalUploadResumeSupported(_ int64) bool {
+	return true
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported
 func (*OsFs) IsAtomicUploadSupported() bool {
 	return true

internal/vfs/s3fs.go

@@ -283,10 +283,28 @@ func (fs *S3Fs) Create(name string, flag, checks int) (File, *PipeWriter, func()
 		})
 		r.CloseWithError(err) //nolint:errcheck
 		p.Done(err)
-		fsLog(fs, logger.LevelDebug, "upload completed, path: %q, acl: %q, readed bytes: %v, err: %+v",
+		fsLog(fs, logger.LevelDebug, "upload completed, path: %q, acl: %q, readed bytes: %d, err: %+v",
 			name, fs.config.ACL, r.GetReadedBytes(), err)
 		metric.S3TransferCompleted(r.GetReadedBytes(), 0, err)
 	}()
+
+	if checks&CheckResume != 0 {
+		readCh := make(chan error, 1)
+
+		go func() {
+			err = fs.downloadToWriter(name, p)
+			readCh <- err
+		}()
+
+		err = <-readCh
+		if err != nil {
+			cancelFn()
+			p.Close()
+			fsLog(fs, logger.LevelDebug, "download before resume failed, writer closed and read cancelled")
+			return nil, nil, nil, err
+		}
+	}
+
 	return nil, p, cancelFn, nil
 }
 
@@ -460,6 +478,12 @@ func (*S3Fs) IsUploadResumeSupported() bool {
 	return false
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (*S3Fs) IsConditionalUploadResumeSupported(size int64) bool {
+	return size <= resumeMaxSize
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported.
 // S3 uploads are already atomic, we don't need to upload to a temporary
 // file
@@ -1026,6 +1050,31 @@ func (*S3Fs) GetAvailableDiskSize(_ string) (*sftp.StatVFS, error) {
 	return nil, ErrStorageSizeUnavailable
 }
 
+func (fs *S3Fs) downloadToWriter(name string, w *PipeWriter) error {
+	fsLog(fs, logger.LevelDebug, "starting download before resuming upload, path %q", name)
+	ctx, cancelFn := context.WithTimeout(context.Background(), preResumeTimeout)
+	defer cancelFn()
+
+	downloader := manager.NewDownloader(fs.svc, func(d *manager.Downloader) {
+		d.Concurrency = fs.config.DownloadConcurrency
+		d.PartSize = fs.config.DownloadPartSize
+		if fs.config.DownloadPartMaxTime > 0 {
+			d.ClientOptions = append(d.ClientOptions, func(o *s3.Options) {
+				o.HTTPClient = getAWSHTTPClient(fs.config.DownloadPartMaxTime, 100*time.Millisecond)
+			})
+		}
+	})
+
+	n, err := downloader.Download(ctx, w, &s3.GetObjectInput{
+		Bucket: aws.String(fs.config.Bucket),
+		Key:    aws.String(name),
+	})
+	fsLog(fs, logger.LevelDebug, "download before resuming upload completed, path %q size: %d, err: %+v",
+		name, n, err)
+	metric.S3TransferCompleted(n, 1, err)
+	return err
+}
+
 func (fs *S3Fs) getStorageID() string {
 	if fs.config.Endpoint != "" {
 		if !strings.HasSuffix(fs.config.Endpoint, "/") {

internal/vfs/sftpfs.go

@@ -540,6 +540,12 @@ func (fs *SFTPFs) IsUploadResumeSupported() bool {
 	return fs.config.BufferSize == 0
 }
 
+// IsConditionalUploadResumeSupported returns if resuming uploads is supported
+// for the specified size
+func (fs *SFTPFs) IsConditionalUploadResumeSupported(_ int64) bool {
+	return fs.IsUploadResumeSupported()
+}
+
 // IsAtomicUploadSupported returns true if atomic upload is supported.
 func (fs *SFTPFs) IsAtomicUploadSupported() bool {
 	return fs.config.BufferSize == 0

internal/vfs/vfs.go

@@ -40,15 +40,17 @@ import (
 )
 
 const (
-	dirMimeType  = "inode/directory"
-	s3fsName     = "S3Fs"
-	gcsfsName    = "GCSFs"
-	azBlobFsName = "AzureBlobFs"
+	dirMimeType      = "inode/directory"
+	s3fsName         = "S3Fs"
+	gcsfsName        = "GCSFs"
+	azBlobFsName     = "AzureBlobFs"
+	preResumeTimeout = 90 * time.Second
 )
 
 // Additional checks for files
 const (
 	CheckParentDir = 1
+	CheckResume    = 2
 )
 
 var (
@@ -62,6 +64,7 @@ var (
 	allowSelfConnections int
 	renameMode           int
 	readMetadata         int
+	resumeMaxSize        int64
 )
 
 // SetAllowSelfConnections sets the desired behaviour for self connections
@@ -94,6 +97,12 @@ func SetReadMetadataMode(val int) {
 	readMetadata = val
 }
 
+// SetResumeMaxSize sets the max size allowed for resuming uploads for backends
+// with immutable objects
+func SetResumeMaxSize(val int64) {
+	resumeMaxSize = val
+}
+
 // Fs defines the interface for filesystem backends
 type Fs interface {
 	Name() string
@@ -113,6 +122,7 @@ type Fs interface {
 	ReadDir(dirname string) ([]os.FileInfo, error)
 	Readlink(name string) (string, error)
 	IsUploadResumeSupported() bool
+	IsConditionalUploadResumeSupported(size int64) bool
 	IsAtomicUploadSupported() bool
 	CheckRootPath(username string, uid int, gid int) bool
 	ResolvePath(virtualPath string) (string, error)
@@ -893,6 +903,14 @@ func SetPathPermissions(fs Fs, path string, uid int, gid int) {
 	}
 }
 
+// IsUploadResumeSupported returns true if resuming uploads is supported
+func IsUploadResumeSupported(fs Fs, size int64) bool {
+	if fs.IsUploadResumeSupported() {
+		return true
+	}
+	return fs.IsConditionalUploadResumeSupported(size)
+}
+
 func updateFileInfoModTime(storageID, objectPath string, info *FileInfo) (*FileInfo, error) {
 	if !plugin.Handler.HasMetadater() {
 		return info, nil

internal/webdavd/handler.go

@@ -215,7 +215,7 @@ func (c *Connection) handleUploadToNewFile(fs vfs.Fs, resolvedPath, filePath, re
 		c.Log(logger.LevelDebug, "upload for file %q denied by pre action: %v", requestPath, err)
 		return nil, c.GetPermissionDeniedError()
 	}
-	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.GetCreateChecks(requestPath, true))
+	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.GetCreateChecks(requestPath, true, false))
 	if err != nil {
 		c.Log(logger.LevelError, "error creating file %q: %+v", resolvedPath, err)
 		return nil, c.GetFsError(fs, err)
@@ -262,7 +262,7 @@ func (c *Connection) handleUploadToExistingFile(fs vfs.Fs, resolvedPath, filePat
 		}
 	}
 
-	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.GetCreateChecks(requestPath, false))
+	file, w, cancelFn, err := fs.Create(filePath, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, c.GetCreateChecks(requestPath, false, false))
 	if err != nil {
 		c.Log(logger.LevelError, "error creating file %q: %+v", resolvedPath, err)
 		return nil, c.GetFsError(fs, err)

sftpgo.json

@@ -9,6 +9,7 @@
     },
     "setstat_mode": 0,
     "rename_mode": 0,
+    "resume_max_size": 0,
     "temp_path": "",
     "proxy_protocol": 0,
     "proxy_allowed": [],