Skip to main content

2607.03.053

Version / Release: 2607.03.053​

Release date: 1 September 2026

Improvements​

Keyboard layout for streaming in the browser​

Several improvements to the keyboard layout that apps use in the browser:

  • Danish layout added: the usage settings of an app configuration now offer "da-DK Qwerty" under "Keyboard layout selected on the remote system". The Danish special characters æ ø å and the characters on AltGr arrive correctly.
  • "Other layout" works as intended: the option previously named "Different layout (use remote system setting)" no longer falls back to the US layout. Characters are now transferred independently of the layout, so they also arrive correctly on remote systems whose layout is not in the list — including characters beyond Western European alphabets such as ą ć ę, ğ ı ş, Greek and Cyrillic characters, and the Euro sign. Shortcuts such as Ctrl+C, Ctrl+V, and Alt+Tab work in this mode as well.
  • Help text for the setting: a help icon next to the setting explains that the layout of the remote system cannot be detected automatically, and which entry to choose. The help text is available in the app configuration and on the app tile in oneclick™ Desk.

A layout from the list transfers all characters reliably, so choose "Other layout" only when the layout of the remote system is unknown or not in the list. In that mode, letter shortcuts such as Ctrl+Z can reach a different letter if the remote system uses a Qwertz-based layout that oneclick™ does not offer. As before, the setting applies to apps that are opened in the browser via oneclick™ Streaming, not to apps opened through the oneclick™ Gateway.

Maintenance mode now takes effect in load balancing pools​

The "Enable maintenance mode" dialog promises that no new logins are allowed on the destinations. In pools with "Sticky user sessions" that promise did not hold: users who had worked on a destination before were connected straight back to it after it went into maintenance, while users without such a binding were distributed correctly. The same pool behaved differently from user to user, and nothing in the admin area showed it.

  • Users whose reuse entry points at a destination in maintenance are now routed to a free destination of the pool instead — silently, without an error message and without anything for them to do.
  • Enabling maintenance mode now releases the reuse entries on the affected destinations, so you no longer have to release them by hand in the pool details. Entries are kept for users in groups that are granted access during maintenance mode, and for users whose session is still running on that destination.
  • Users with a running session can still start further apps on that destination. Moving them while the session is live could damage their Windows profile or leave them with a temporary one. These users now receive a notification asking them to save their documents and sign out of the system.
  • The dialog explains what this means and what to do about it, and the "Show sessions" link takes you to the details of the affected destination pool. The link appears only for pools that reuse existing user sessions.
  • If every destination of the pool is in maintenance, users now see the maintenance notice instead of a message about missing free resources.

App tiles show pool maintenance before the click​

When an app uses "Best destination of the load balancing pool" as its destination, its tile is now greyed out as soon as every destination of the pool is in maintenance for that user, and shows the maintenance notice on hover. Previously the tile looked normal, and users only found out after clicking and waiting for the start to fail. Tiles change state while users are signed in, so nobody has to sign in again.

  • If only part of the pool is in maintenance, the tile stays available and the app starts on a free destination — unchanged.
  • The tooltip shows your notice when every destination of the pool carries the same one. If the notices differ or none is set, users see a neutral text instead, because a maintenance time taken from one destination could mislead about another.
  • Tiles now apply the same rules as the app start itself. With a specific destination, users of the oneclick™ administration division and cross-division support with "Allow support access during maintenance" now see a usable tile, because the start lets them through anyway.

Group detail view opens without loading every user​

Opening a group in the admin area no longer loads the complete user list of the division first. The "Users" tile now requests only the members of that group and appears while the rest of the page is still loading — previously it waited for the entire list to arrive page by page, which is what made large divisions slow. Emptying a group now shows the empty state immediately instead of leaving the tile on a loading animation. Loading the user list itself is quicker as well, because a check that ran on every page no longer searches data of the whole platform.

Bug Fixes​

  • Filter requests to the oneclick™ API apply all filters again: Requests to the /filter endpoints now evaluate every filter that is sent — OR combinations, several filters within one filter group, and several filter groups — and they apply sort, limit, start, and page again. Previously only the first filter of the first group took effect, so a request returned more records than expected without reporting an error. Excluding filters (not) match correctly again as well, and a malformed filter or sort parameter now returns a clear error instead of an empty list. The includes parameter is unaffected: it was removed deliberately and this fix does not restore it.
  • Login via an identity provider recovers automatically: After a service interruption, login via an identity provider (Single Sign-on) could keep failing until the platform was restarted manually. oneclick™ now detects this state itself and corrects it automatically.
  • Networks with an overlapping address range can be created again for cloud connections: For AWS, Azure, Google Cloud, Open Telekom Cloud, Ionos, Exoscale, Vultr, Orange, Huawei, and Ventus, a second network with an overlapping CIDR range in the same connection is accepted again — these Cloud Providers allow identical address space within one cloud account. The check stays in place for STACKIT and IPsec connections, where an overlap does cause a conflict. Networks and subnets that failed to be created or have already been deleted no longer occupy their address range, so you can create a failed network again with the same range. The check now also applies when you edit the CIDR range of a network or add a subnet.
  • AD/LDAP sync no longer proposes users with a second connection for deletion: Users who are still synchronized through a second AD/LDAP connection no longer appear in the list of users proposed for deletion during sync.
  • AD/Azure import no longer aborts entirely when one user hangs: A single user whose processing used to stall no longer aborts the whole import — it continues for the remaining users.
  • AD/LDAP credentials no longer appear in API responses: The AD/LDAP connection's username and password are no longer returned in the corresponding API responses.
  • STACKIT and OpenStack connections report the actual error: Errors from STACKIT and OpenStack connections now show the provider's actual message instead of a generic one, and a missing response now results in a timeout instead of an indefinite wait. If the multi-project query gets no response at all, it is now reported as a clear error instead of silently returning an empty list.
  • Open Telekom Cloud pricing is now complete: Instances of the ecsflex family are now shown with a price as well.
  • AD sync group display corrected: The group display in AD sync now uses the correct separator throughout and shows the right label for empty groups.