v0.23.0

Added

• Platform administrators can now set how long deleted orders and discarded
  drafts are kept. After this period, they and their history are permanently
  deleted. When a retention period is set, each deleted order’s page shows the
  days remaining and the deadline for restoring it. By default, deleted orders
  and discarded drafts are kept indefinitely.
• Service owners can now define a namespace name template so orders deploy to
  the namespace the chart actually creates. The order form previews the name as
  the required fields are filled in.
• Added “Restore order” to the order list and the deleted order’s page. It
  creates a new draft with the order’s parameters without restoring service
  data.
• Added the SHUTDOWN_DRAIN environment variable to help prevent interrupted
  requests during Kubernetes rollouts. During shutdown, the portal first reports
  that it is not ready to accept requests, then continues serving requests
  already heading to it for the configured duration. The default is 0, with no
  delay.
• Added the “Argo CD project” field to stand settings. For stands in
  “Deployment” mode, the stand page checks that the project exists and that the
  portal account can read and sync its applications.

Changed

• Organized the Admin sidebar into groups for monitoring, configuration, and
  maintenance. Shortened the menu labels to “Orders”, “Publications”, and
  “Status”. When the menu is collapsed, separators replace group headings.
• Replaced “Copy to stand” with “Create copy” in the orders list and on the
  order page. You can now create a draft on the same stand with a new service
  name or copy an order to another stand.

Fixed

• The same service ordered on different stands now gets distinct application
  names, preventing conflicts in Argo CD. Application names for existing orders
  remain unchanged. If an application name is already used by an order on
  another stand in the same Argo CD instance, the portal rejects the new order
  and asks you to choose a different service name.
• Removed “Sync with cluster” from the order list row menu and the order page
  actions menu for administrators viewing deleted orders. The portal also
  rejects synchronization requests for deleted orders and drafts, even if sent
  outside the interface.
• The order form no longer blocks submission when required parameters have
  service defaults, including parameters hidden from the form. Default values
  appear as hints, with dropdowns and checkboxes set to their default states.
  Required parameters without defaults still need to be filled in.
• Fixed several issues affecting portals running with multiple instances. “Check
  again” on the “Configuration” page now runs checks on every instance so the
  page no longer shows stale results from another instance. After a leader
  switch, administrators no longer receive duplicate configuration alerts, such
  as warnings about an expiring token or a disabled webhook, and receive alerts
  again if a resolved problem returns. Order and user counts in monitoring are
  no longer inflated when summed across instances after a leader switch. During
  shutdown, the leader now finishes its current background order processing
  before handing the role to another instance.