* Removing timeout from Webdav
The timeout will be controlled externally by the TimeoutObservingStream and or the cancellation token.
Timeout for List set to default of 10 minutes
Since filesystems may take quite sometime to list large number of files, the timeout was set to a high value.
Previous timeout of 30 seconds was triggering in a case reported by user.
Timeout for CreateFolder and DeleteAsync was kept to default of 30 seconds.
Minor formmating/refactoring was done.
* Removing timeout from Webdav
The timeout will be controlled externally by the TimeoutObservingStream and or the cancellation token.
Timeout for List set to default of 10 minutes
Since filesystems may take quite sometime to list large number of files, the timeout was set to a high value.
Previous timeout of 30 seconds was triggering in a case reported by user.
Timeout for CreateFolder and DeleteAsync was kept to default of 30 seconds.
Minor formating/refactoring was done.
* Fixes to Integrated Authentication
This is a fix to the webdav when using integrated authentication, the hint for the underlying issue was found on an old [thread] (https://github.com/dotnet/wcf/issues/2923). By setting up a vm environment with Windows 2022 Server+IIS/Webdav it was possible to reproduce the [issue](https://github.com/duplicati/duplicati/issues/5872) and fix it.
Change to validation script is to avoid the requirement of filling the username when integrated authentication is true.
Modernize HTTP client implementation across backend services by:
- Replace legacy HttpWebClient with modern HttpClient
- Implement proper async/await patterns throughout the codebase
- Add cancellation token support for improved request handling
- Refactor related code for better maintainability
This change is part of our broader initiative to modernize the codebase
and improve its resilience through proper asynchronous operations.
* Fixed nullability warnings in `Utility.cs`
* Fixed a number of nullability issues and warnings
* More nullable fixes
* Fixed the overwrite issue
* Fixed additional nullability issues
* Removed additional nullability issues
* More nullability fixes and added some xmldoc
* Added more xmldocs and removed an unused argument
* Removed more warnings
Adds support for user-configurable read and write buffer sizes while enforcing protocol-negotiated maximums. When not specified, operations default to using the maximum sizes negotiated during the SMB session setup.
Tests revealed that by not using a single connection throttling issues appeared when connecting to Windows shares (possibly due to multiple authentications in a short period of time)
The 'no-reuse-connection' option controls whether SMB/CIFS connections are reused:
When enabled, creates a new connection and authentication for each file operation
When not specified, defaults to reusing the same connection across operations
* Primary functionality implemented for CIFS
This will add CIFS support to Duplicati.
One of the advantages is that the resource does not need to be mounted as the backend connects and authenticates directly to the resource via TCP/or Netbios over TCP.
The authentication supported is domain/username/password. At present integrated Windows authentication is not supported.
This implementation relies on SMBLibrary library (https://github.com/TalAloni/SMBLibrary).
* Fixes to XML documentation
* Implementing single CIFS connection + Fix on PutAsync
Single CIFS connection to avoid re-authentication on every operation.
Fix to PutAsync which only because apparent after file sizes were larger than _smb2Client.MaxWriteSize
* Adding CI tests for CIFS
The CI test for CIFS is self contained, requiring no environment variables. Also updated README.
This introduces a few constants in favor of multiple strings with the same value.
The supported arguments in the module have been augmented to better report possible values and default values.
The options parser has been updated to work better with CLI where the options are not default filled.
* Added options to insert delays in backend tester
* Added wait delays to Azure and GoogleDrive to see it the CI is more stable with small delays
* Added more debug output to the backend tester
* Fixed dropbox reporting OAuth quota errors, when it is likely Dropbox API quota errors
* Implemented exception detection on AzureBlob backend
* Also showing sleep times in output
* Fixed a bug with the dropbox quota response
* Fixed error message to report correct number of files
* Fixed suppressed warning
* Change displayname of OneDrive v2
Since OneDrive v1 is removed, there is no need to name the current backend `v2`.
* Renamed everything but the protocol key
* Fixed issues with timeout in msgraph
Fixed a place where the cancellation token was not applied.
Fixed a problem with msgraph not sending correct DNS information.
Fixed a problem with msgraph downloading the entire file before streaming it, bypassing throttle and tripping the timeout.
Added `ITimeoutExemptBackend` to selectively disable timeouts for backends that do not support it.
Updated the backendmanager to support `ITimeoutExemptBackend`.
Updated the backendmanager to abort as soon as a timeout happens.
* Updated BackendTester to use the `TimeoutObservingStream` and support the `--read-write-timeout` setting.
Updated backend tester code to use `var`
* Fixed a bug reading the `read-write-timeout`
* Updated to using constants with names instead of `-1`.
- Add support for subfolders
- Implement robust file selection using fileID for Download/Delete operations instead of concatenation of path + filename
- Improved validation for folder existence during backend connectivity tests supporting subfolders.
This issue does not affect the pCloud backend per se and its functioning as expected.
The issue is related to a symptom the BackendTester revealed as discussed here:
https://github.com/duplicati/duplicati/pull/5709
It turns out the backends should assume the caller has set the input stream at the correct position and take it from there.
Therefore the fix is simply to remove the .Position = 0; which was well intended as a way to make sure the stream was at the beginning, but thats not the pattern to follow when integrating backends.
A future blog post/tutorial on new backend integrations will mention these premises.
Changed the strategy to use absolute paths for all operations, to avoid cases where a silent reconnect would end up in a different folder.
Add backwards compatible option to use CWD style relative operations.
Added support for detecting the SSL certificate hash when there is a certificate error.
Kenneth Skovhede