STAGE Decommissioning
π€ Generated with Claude Code
β οΈ Not yet reviewed
Co-Authored-By: Claude (noreply@anthropic.com)
The decision to retire the STAGE environment is made. Nothing has been removed yet: STAGE is still up and still deployed to, and the timing is not ours alone to set β three partner insurers use it, and each needs a resolution first.
This document records where that stands and what removal will involve. It becomes the decommissioning record once the work starts, in the shape of vudev-decommissioning.md.
| Field | Value |
|---|---|
| Environment | STAGE / Stage (ASPNETCORE_ENVIRONMENT=Stage) |
| Status | Decision made, not started |
| Target | End of 2026 β a goal, not a commitment; see below |
| Key Vault | at-tgca-tis-s-default (Together-Fachtest subscription, TIS-Secrets) |
| Enterprise App / Group | stage-apps / Stage-Applications |
| Host | stage.servicebytogether.at |
| API gateway | api-stage.togethercca.com |
| Deployment group | stage.servicebytogether.at |
| Repository survey | 2026-08-25, across all repos under the source root |
What has to happen firstβ
The blockers are external, which is why the target date is soft.
| Stakeholder | Where it stands |
|---|---|
| Generali | ready to tackle the topic in autumn |
| VIG | needs a solution for the yearly Klappung |
| UNIQA | not yet discussed |
Until all three have a replacement for whatever they use STAGE for, the environment stays. Treat end of 2026 as an intention: it depends on partners we do not control.
What removal will involveβ
Nothing here is done. The list is taken from what VUDEV actually needed, which is the only proven checklist we have β the lesson from that retirement being that the Azure and server side is quick and the configuration cleanup is the long tail.
| Item | Status |
|---|---|
| Deployment pipelines and release stages targeting STAGE removed | β |
Deployment group stage.servicebytogether.at and its agents removed | β |
Key Vault at-tgca-tis-s-default deleted | β |
Enterprise App stage-apps and group Stage-Applications deleted | β |
| Servers offline | β |
DNS records stage.servicebytogether.at and api-stage.togethercca.com retired | β |
| Gateway route configuration for the STAGE context removed | β |
| Documentation in this repository updated | β |
| Stale configuration removed from the application repositories | β |
Scale of the configuration cleanupβ
Baseline at survey time (2026-08-25), excluding build output:
| Count | |
|---|---|
appsettings.stage*.json and nlog.stage*.config files | 41, across 26 repositories |
Configuration files referencing the Key Vault at-tgca-tis-s-default | 12, across 12 repositories |
The repositories carrying the most are tis.volltextsuche (6), tis.gf.smr5 (4) and
tis.gf.esign (3); the remaining 23 hold one or two each.
As with VUDEV, an AKV reference to a deleted vault has no fallback β an application
started with ASPNETCORE_ENVIRONMENT=Stage after the vault is gone fails at start-up rather
than degrading. Deleting the vault before the configuration is cleaned up is therefore safe
only because nothing should still be starting in that environment; verify that rather than
assume it.
Unlike VUDEV, STAGE is a first-class environment in the platform library:
TisEnvironments.Stage and IsTisStage() in Tis.Hosting.Extensions. IsTisStage() also
matches .NET's standard Staging, and it is used by consuming services β so removing it is a
breaking API change for the NuGet package, not a cleanup. That decision is separate from
retiring the servers and SHOULD be taken on its own merits.
Re-checking progressβ
From the source root (all repositories side by side), excluding build output:
# files named after the environment (case-insensitive)
rg --files --glob-case-insensitive -g '*stage*' -g '!**/bin/**' -g '!**/obj/**'
# references to the Key Vault
rg -l 'at-tgca-tis-s-default' -g '!**/bin/**' -g '!**/obj/**'
# host names
rg -l 'stage\.servicebytogether\.at|api-stage\.togethercca\.com' -g '!**/bin/**' -g '!**/obj/**'
Related: together-on-premises.md for what the environment consists of, and ../secrets/README.md for the vault and service principal.