If you prefer to auto-create the first admin you can enable the
"create_default_admin" configuration key and SFTPGo will work as before.
You can also create the first admin by loading initial data: now you can
set both username and password, before you could only change the password
The actions to run synchronously can be configured via the `execute_sync`
configuration key.
Executing an action synchronously means that SFTPGo will not return a result
code to the client until your hook have completed its execution.
Fixes#409
X-Forwarded-For, X-Real-IP and X-Forwarded-Proto headers will be ignored
for hosts not included in this list.
This is a backward incompatible change, before the proxy headers were
always used
argon2id has a high memory cost and, if not properly tuned, it can lead to
resource starvation.
Advanced users can still configure and use argon2id.
Passwords stored as argon2id will continue to work
Enabling the "skip_natural_keys_validation" data provider setting,
the natural keys for REST API/Web Admin as usernames, admin names,
folder names are not restricted to unreserved URI chars
Fixes#334#308
For each user you can now configure:
- TLS certificate auth
- TLS certificate auth and password
- Password auth
For TLS auth, the certificate common name must match the name provided
using the "USER" FTP command
ioutil is deprecated in Go 1.16 and SFTPGo is an application, not
a library, we have no reason to keep compatibility with old Go
versions.
Go 1.16 fix some cifs related issues too.
- add JWT authentication
- admins are now stored inside the data provider
- admin access can be restricted based on the source IP: both proxy
header and connection IP are checked
- deprecate REST API CLI: it is not relevant anymore
Some other changes to the REST API can still happen before releasing
SFTPGo 2.0.0
Fixes#197
- allow to disable active mode
- allow to disable SITE commands
- add optional support for calculating hash value of files
- add optional support for the non standard COMB command
Config name is a Viper concept used for searching a specific file
in various paths with various extensions.
Making it configurable is usually not a useful feature
as users mostly want to define a full or relative path
to a config file.
This change replaces config name with config file.
When true, users' Google Cloud Storage credentials will be written to
the data provider instead of disk.
Pre-existing credentials on disk will be used as a fallback
Fixes#201
The common package defines the interfaces that a protocol must implement
and contain code that can be shared among supported protocols.
This way should be easier to support new protocols
host_key defines the private host keys as plain list of strings.
Remove the other deprecated config params from the default config too.
Signed-off-by: Nicola Murino <nicola.murino@gmail.com>
HTTP clients are used for executing hooks such as the ones used for custom
actions, external authentication and pre-login user modifications.
This allows, for example, to use self-signed certificate without defeating the
purpose of using TLS
"proxy_allowed" setting allows to specify the allowed IP address and IP
ranges that can send the proxy header. This setting combined with
"proxy_protocol" allows to ignore the header or to reject connections
that send the proxy header from a non listed IP
A custom program can be executed before the users login to modify the
configurations for the user trying to login.
You can, for example, allow login based on time range.
Fixes#77
md5sum, sha1sum are used by rclone.
cd, pwd improve the support for RemoteFiles mobile app.
These commands are all implemented inside SFTPGo so they work even
if the matching system commands are not available, for example on Windows
we support resume only if the client sets the correct offset while resuming
the upload.
Based on the specs the offset is optional for resume, but all the tested
clients sets a right offset.
If an invalid offset is given we interrupt the transfer with the error
"Invalid write offset ..."
See https://github.com/pkg/sftp/issues/295
This commit add a new upload mode: "atomic with resume support", this acts
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
The update is atomic so no transaction is needed.
Addionally a transaction will ask for a new connection to the pool
and this can deadlock if the pool has a max connection limit too low.
Also make configurable the pool size instead of hard code to the cpu number.
Fixes#47
