v2.19.0
Changelog for reva 2.19.0 (2024-02-22)
The following sections list the changes in reva 2.19.0 relevant to
reva users. The changes are ordered by importance.
Summary
- Fix #4464: Do not lose revisions when restoring the first revision
- Fix #4516: The sharemanager can now reject grants with resharing permissions
- Fix #4512: Bump dependencies
- Fix #4481: Distinguish failure and node metadata reversal
- Fix #4456: Do not lose revisions when restoring the first revision
- Fix #4472: Fix concurrent access to a map
- Fix #4457: Fix concurrent map access in sharecache
- Fix #4498: Fix Content-Disposition header in dav
- Fix #4461: CORS handling for WebDAV requests fixed
- Fix #4462: Prevent setting container specific permissions on files
- Fix #4479: Fix creating documents in the approvider
- Fix #4474: Make /dav/meta consistent
- Fix #4446: Disallow to delete a file during the processing
- Fix #4517: Fix duplicated items in the sharejail root
- Fix #4473: Decomposedfs now correctly lists sessions
- Fix #4528: Respect IfNotExist option when uploading in cs3 metadata storage
- Fix #4503: Fix an error when move
- Fix #4466: Fix natsjskv store
- Fix #4533: Fix recursive trashcan purge
- Fix #4492: Fix the resource name
- Fix #4463: Fix the resource name
- Fix #4448: Fix truncating existing files
- Fix #4434: Fix the upload postprocessing
- Fix #4469: Handle interrupted uploads
- Fix #4532: Jsoncs3 cache fixes
- Fix #4449: Keep failed processing status
- Fix #4529: We aligned some OCS return codes with oc10
- Fix #4507: Make tusd CORS headers configurable
- Fix #4452: More efficient share jail
- Fix #4476: No need to unmark postprocessing when it was not started
- Fix #4454: Skip unnecessary share retrieval
- Fix #4527: Unify datagateway method handling
- Fix #4530: Drop unnecessary grant exists check
- Fix #4475: Upload session specific processing flag
- Enh #4501: Allow sending multiple user ids in one sse event
- Enh #4485: Modify the concurrency default
- Enh #4526: Configurable s3 put options
- Enh #4453: Disable the password policy
- Enh #4477: Extend ResumePostprocessing event
- Enh #4491: Add filename incrementor for secret filedrops
- Enh #4490: Lazy initialize public share manager
- Enh #4494: Start implementation of a plain posix storage driver
- Enh #4502: Add spaceindex.AddAll()
Details
- Bugfix #4464: Do not lose revisions when restoring the first revision
We no longer prevent modifying grants on locked resources.
- Bugfix #4516: The sharemanager can now reject grants with resharing permissions
When disabling resharing we also need to prevent grants from allowing any grant permissions.
- Bugfix #4512: Bump dependencies
We updated the dependencies to at least match the ones used in ocis.
- Bugfix #4481: Distinguish failure and node metadata reversal
When the final blob move fails we must not remove the node metadata to be able to restart the
postprocessing process.
- Bugfix #4456: Do not lose revisions when restoring the first revision
We fixed a problem where restoring the very first version of a file could delete the current
version.
- Bugfix #4472: Fix concurrent access to a map
We fixed the race condition that led to concurrent map access in a publicshare manager.
- Bugfix #4457: Fix concurrent map access in sharecache
We fixed a problem where the sharecache map would sometimes cause a panic when being accessed
concurrently.
- Bugfix #4498: Fix Content-Disposition header in dav
We have added missing quotes to the Content-Disposition header in the dav service. This fixes
an issue with files containing special characters in their names.
- Bugfix #4461: CORS handling for WebDAV requests fixed
We now correctly handle CORS headers for WebDAV requests.
- Bugfix #4462: Prevent setting container specific permissions on files
It was possible to set the 'CreateContainer', 'Move' or 'Delete' permissions on file
resources with a CreateShare request. These permissions are meant to be only set on container
resources. The UpdateShare request already has a similar check.
- Bugfix #4479: Fix creating documents in the approvider
We fixed a problem with the approvider where an error was reported to the user even though the
file was created properly.
- Bugfix #4474: Make /dav/meta consistent
We now also return absolute paths for shares in the share jail in the /dav/meta endpoint.
- Bugfix #4446: Disallow to delete a file during the processing
We want to disallow deleting a file during the processing to prevent collecting the orphan
uploads.
- Bugfix #4517: Fix duplicated items in the sharejail root
We fixed a bug, that caused duplicate items to listed in the sharejail, when a user received
multiple shares for the same resource.
- Bugfix #4473: Decomposedfs now correctly lists sessions
- Bugfix #4528: Respect IfNotExist option when uploading in cs3 metadata storage
- Bugfix #4503: Fix an error when move
We fixed a bug that caused Internal Server Error when move using destination id
- Bugfix #4466: Fix natsjskv store
Small mistake in last iteration made authorization not working.
- Bugfix #4533: Fix recursive trashcan purge
We have fixed a bug in the trashcan purge process that did not delete folder structures
recursively.
- Bugfix #4492: Fix the resource name
We fixed a problem where after renaming resource as sharer the receiver see a new name for dav and
wedav endpoints.
- Bugfix #4463: Fix the resource name
We fixed a problem where after renaming resource as sharer the receiver see a new name.
- Bugfix #4448: Fix truncating existing files
We fixed a problem where existing files kept their content when being overwritten by a 0-byte
file.
- Bugfix #4434: Fix the upload postprocessing
We fixed the upload postprocessing when the destination file does not exist anymore.
- Bugfix #4469: Handle interrupted uploads
We fixed a bug where interrupted uploads were not discarded properly.
- Bugfix #4532: Jsoncs3 cache fixes
The jsoncs3 share manager now retries persisting if the file already existed and picks up the
etag of the upload response in all cases.
- Bugfix #4449: Keep failed processing status
We now keep tho postprocessing status when a blob could not be copied to the blobstore.
- Bugfix #4529: We aligned some OCS return codes with oc10
- Bugfix #4507: Make tusd CORS headers configurable
We bumped tusd to 1.13.0 and made CORS headers configurable via mapstructure.
- Bugfix #4452: More efficient share jail
The share jail was stating every shared recource twice when listing the share jail root. For no
good reason. And it was not sending filters when it could.
- Bugfix #4476: No need to unmark postprocessing when it was not started
- Bugfix #4454: Skip unnecessary share retrieval
- Bugfix #4527: Unify datagateway method handling
The datagateway now unpacks and forwards all HTTP methods
- Bugfix #4530: Drop unnecessary grant exists check
At least the jsoncs3 share manager properly returns an ALREADY_EXISTS response when trying to
add a share to a resource that has already been shared with the grantee.
- Bugfix #4475: Upload session specific processing flag
To make every upload session have a dedicated processing status, upload sessions are now
treated as in processing when all bytes have been received instead of checking the node
metadata.
- Enhancement #4501: Allow sending multiple user ids in one sse event
Sending multiple user ids in one sse event is now possible which reduces the number of sent
events.
- Enhancement #4485: Modify the concurrency default
We have changed the default MaxConcurrency value from 100 to 5 to prevent too frequent gc runs on
low memory systems.
- Enhancement #4526: Configurable s3 put options
The s3ng blobstore can now be configured with several options:
s3.disable_content_sha254, s3.disable_multipart, s3.send_content_md5,
s3.concurrent_stream_parts, s3.num_threads and s3.part_size. If unset we default
to s3.send_content_md5: true, which was hardcoded before. We also default to
s3.concurrent_stream_parts: true and s3.num_threads: 4 to allow concurrent uploads
even when s3.send_content_md5 is set to true. When tweaking the uploads try setting
s3.send_content_md5: false and s3.concurrent_stream_parts: false first, as this will
try to concurrently stream an uploaded file to the s3 store without cutting it into parts first.
- Enhancement #4453: Disable the password policy
We reworked and moved disabling the password policy logic to the ocis.
- Enhancement #4477: Extend ResumePostprocessing event
Instead of just sending an uploadID, one can set a postprocessing step now to restart all
uploads in this step Also adds a new postprocessing step - "finished" - which means that
postprocessing is finished but the storage provider hasn't acknowledged it yet.
- Enhancement #4491: Add filename incrementor for secret filedrops
We have added a function that appends a number to the filename if the file already exists in a
secret filedrop. This is useful if you want to upload a file with the same name multiple times.
- Enhancement #4490: Lazy initialize public share manager
Unlike the share manager the public share manager was initializing its data structure on
startup. This can lead to failed ocis starts (in single binary case) or to restarting sharing
pods when running in containerized environment.
- Enhancement #4494: Start implementation of a plain posix storage driver
We started to lay the groundwork for a new posixfs storage driver based on decomposedfs. We also
refactored decomposedfs to be a bit more modular and cleaned up the initialization.
- Enhancement #4502: Add spaceindex.AddAll()
We now expose an AddAll() function that allows adding multiple entries to a space index with a
single lock.