26.10.2 Maintenance Patch
Changes compared to 26.10.1
This is a maintenance patch to the 26.10.0 Quarterly Release. See 26.10.0 Quarterly Release for what's new in 26.10.
Behavior changes to know about
- Microsoft 365 calendars: Calendar backups now default to the calendarView/delta API, with an automatic page-size workaround for a Microsoft Graph pagination bug, and fall back to the events API when a calendar still cannot be listed. Calendars with large numbers of recurring events that previously failed to back up can now complete. Calendar events that lie entirely before October 2020 or after November 2050 are not included while calendarView/delta succeeds
- Microsoft 365 backup jobs now include unchanged calendar events and contacts in the reported job size, so job sizes may look larger than before
- Software update jobs that have been in a Running status for more than 24 hours are marked as abandoned. No email report is sent for an abandoned job that started more than 7 days ago
- A slow or stalled software update download no longer keeps the update job in a Running status for several hours. Each download attempt is now bounded, and the job fails after a limited number of attempts
Enhancements
Magnus Box Application
- Improved logging by the backup app's background service on Linux: messages are now written to the systemd journal with their severity, or to syslog on systems without systemd (including Synology)
- Improved the security of communication between the backup app's background service processes
Web Portal
- Improved the Bulk Update Campaign to re-send the update request up to three times, at least an hour apart, before marking a device as failed
- Improved the Update Software action: devices running 26.10.2 or later confirm that they accepted the request, and a device that already has an update in progress reports this instead of queuing a second update
- Added Created, Modified and Associated Devices columns (hidden by default) to the Storage Vaults table, Created and Modified columns (hidden by default) to the Protected Items table, and Protected Item ID and Storage Vault ID columns (hidden by default) to the Job Logs table
- Significantly improved the performance of in-browser restores in the Web Portal, particularly for low-latency Storage Vaults
- Significantly improved the performance of job searches on portals with many jobs, including the Recent Activity page, the homepage job widgets and custom job searches, and the job list and job count APIs
- Significantly improved the performance of email reports and email report previews on portals with many jobs
- Reduced the background load of job log pruning, job log compression, retrying failed jobs and software update campaigns on portals with many jobs
Storage Vaults
- Improved the speed of retention passes when removing unreferenced data from Storage Vaults with large indexes
Admin API
- Added the AdminAdminUserSet API, for updating another administrator account's permissions, password and security settings without rewriting the whole configuration
- Added an optional TargetOrgID parameter to the AdminAdminUserDelete and AdminAdminUserList APIs, to delete or list administrators in a specific organization
Bug Fixes
Microsoft 365
- Improved the reliability of calendar backups by defaulting to the calendarView/delta API, as described under Behavior changes to know about. Calendars with large numbers of recurring events that previously failed to back up can now complete
Microsoft SQL Server
- Fixed an issue with Microsoft SQL Server restores reporting success before SQL Server had finished restoring the database, which could leave a large database in the RESTORING state
- Fixed an issue with Microsoft SQL Server backup and restore jobs reporting success when the VDI data transfer was aborted or timed out, or when the RESTORE statement failed
Restores
- Fixed an issue with Disk Image granular restores, and restoring a Disk Image as a VMware or Hyper-V virtual disk, failing with "Access is denied" when the destination folder does not grant access to the backup service account, such as a folder inside a user profile
- Fixed an issue with granular restores of Disk Image, Hyper-V and VMware Protected Items restoring heavily fragmented files from NTFS volumes incompletely while reporting success, and omitting some files in large folders
- Fixed an issue with in-browser restore failing to detect the region of a Wasabi Storage Vault, reporting a CORS error
Storage Vaults and retention
- Fixed an issue where uploads to Magnus Box Cloud Storage could fail with a "Content-Md5 you specified did not match" error for files larger than 16 MiB
- Fixed an issue where uploads to Magnus Box Cloud Storage could fail with a "ContentLength with Body length 0" error for files smaller than 16 MiB after a temporary error from the storage provider
- Fixed an issue with high memory usage during retention passes on Storage Vaults with large indexes
- Fixed an issue with retention passes using on-disk indexes performing a slow full scan of the Storage Vault to measure its size after removing unreferenced data
- Fixed an issue where Storage Vaults requested from an Impossible Cloud (Partner API) (Object Lock) Storage Template were created with Object Lock disabled. Newly requested vaults from that template are created with Object Lock enabled. If you created vaults from this template before the patch, check whether Object Lock is on
- Fixed an issue with the Region field being blank when editing a Storage Vault created from an Impossible Cloud (Partner API) Storage Template in the Web Portal
- Fixed an issue with the Web Portal resetting when saving a Storage Vault created from an Impossible Cloud (Partner API) Storage Template
Web Portal
- Fixed an issue where saving a webhook always used the JSON message format, which showed an "Invalid URL" error for Email and Amazon SNS/EventBridge webhooks and changed existing Slack webhooks to JSON when they were edited. If you edited a Slack webhook, check that it still uses the Slack message format
- Fixed an issue where the page refreshed after cancelling the Protected Item or Storage Vault editor
- Fixed an issue where the Storage Vault dropdown in the schedule editor showed an empty Storage Vaults heading. This affected schedules in policies, and users with no Storage Vaults
- Fixed an issue where the Protected Item ID column on the Recent Activity page was not hidden by default
- Fixed an issue where the job report page could stop loading if the conflicting job could not be looked up
- Fixed an issue where forcing an update of a device's login URL could save an invalid Web Portal address on the device
Magnus Box Application
- Fixed an issue with incorrect times for the Alberta, Northwest Territories and Manitoba (Canada) and Cuba timezones on Windows, caused by an outdated embedded timezone database
- Fixed an issue where a software update job could remain in a Running status indefinitely if the app's update process was interrupted before the installer was launched
- Fixed an issue where Protected Items, Storage Vaults and schedules created in the app recorded a creation time in January 1970
- Fixed an issue where the backup app's background service on Linux could stop, losing any running jobs, after the systemd journal service was stopped and started
- Fixed an issue where unexpected errors in the backup app's background service's dispatcher process were not logged
- Fixed an issue where Linux backups in a memory-limited container could be selected to preferentially run retention passes, because the host's physical RAM was used instead of the container's memory limit
- Fixed an issue with Linux backups of Plesk subscription directories skipping root-owned log files in
/var/www/vhosts/<subscription>/logswith a "permission denied" warning - Fixed an issue where a failure to apply login details pushed from the Web Portal was not logged by the backup app's background service
Admin API
- Fixed an issue where the AdminAdminUserDelete API could not delete organization administrator accounts, and could delete a top-level administrator with the same username instead
- Fixed an issue where the AdminAdminUserNew API reported an internal error instead of a validation error when given an unknown organization ID
Job logs
- Fixed an issue where a single job whose logs could not be compressed stopped job log compression for all other jobs