Merge pull request 'Release v1.14.1 — ansible certs-dir fix + Robokassa go-live' (#239) from development into master

The Ansible base-directory loop created /opt/scrabble/certs at mode 0750
(deploy:deploy), like config/dumps/images. The gateway (and backend) run as
the distroless nonroot UID 65532 — not the deploy user — so the container
cannot traverse a 0750 certs dir and fails at startup with
"mtls: load server keypair: ... permission denied", crash-looping.

This is latent: a long-running container holds the keypair in memory and
never re-reads the file, so the misconfig only bites when a container
restarts (a host reboot / redeploy). A hoster maintenance reboot exposed it
on prod — the gateway came back crash-looping while the deploy could not SSH
in mid-reboot.

Split certs out of the 0750 loop and create it 0755 (traversable). The keys
stay 0644 by design (the gateway compose relies on it); the host is
single-tenant + SSH-access-controlled, so a traversable certs dir adds no
meaningful exposure. The live prod host was already chmod-fixed by hand; this
keeps the next provisioning run from re-tightening it.

.gitea/workflows/prod-deploy.yaml

@@ -99,6 +99,14 @@ jobs:
       GRAFANA_ADMIN_PASSWORD: ${{ secrets.PROD_GRAFANA_ADMIN_PASSWORD }}
       TELEGRAM_BOT_TOKEN: ${{ secrets.PROD_TELEGRAM_BOT_TOKEN }}
       GATEWAY_VK_APP_SECRET: ${{ secrets.GATEWAY_VK_APP_SECRET }}
+      # Robokassa direct-rail (backend BACKEND_ROBOKASSA_*): the prod shop login + the pass phrases
+      # that sign the launch request / verify the Result callback, and the test-mode flag — a var so
+      # go-live is a flag flip, not a secret redeploy ("1" runs test payments against the test
+      # passwords; empty/"0" is live). An empty login leaves the direct rail disabled.
+      ROBOKASSA_MERCHANT_LOGIN: ${{ secrets.PROD_BACKEND_ROBOKASSA_MERCHANT_LOGIN }}
+      ROBOKASSA_PASSWORD1: ${{ secrets.PROD_BACKEND_ROBOKASSA_PASSWORD1 }}
+      ROBOKASSA_PASSWORD2: ${{ secrets.PROD_BACKEND_ROBOKASSA_PASSWORD2 }}
+      ROBOKASSA_TEST: ${{ vars.PROD_BACKEND_ROBOKASSA_TEST }}
       # VK ID web login: the "Web" app id (the gateway reuses it as GATEWAY_VK_ID_APP_ID at
       # runtime) + the app's protected key. Both shared across contours. The redirect URL is
       # derived from PUBLIC_BASE_URL in deploy/write-prod-env.sh.

.gitea/workflows/prod-rollback.yaml

@@ -61,6 +61,12 @@ jobs:
       # the SAME env.sh (email / VK login / Grafana alerts survive a rollback). TELEGRAM_MINIAPP_URL
       # and GRAFANA_ROOT_URL are derived from PUBLIC_BASE_URL in deploy/write-prod-env.sh.
       GATEWAY_VK_APP_SECRET: ${{ secrets.GATEWAY_VK_APP_SECRET }}
+      # Robokassa direct-rail: the rollback re-renders the same runtime env (write-prod-env.sh), so
+      # it must carry the same credentials or the direct rail goes dark after a rollback.
+      ROBOKASSA_MERCHANT_LOGIN: ${{ secrets.PROD_BACKEND_ROBOKASSA_MERCHANT_LOGIN }}
+      ROBOKASSA_PASSWORD1: ${{ secrets.PROD_BACKEND_ROBOKASSA_PASSWORD1 }}
+      ROBOKASSA_PASSWORD2: ${{ secrets.PROD_BACKEND_ROBOKASSA_PASSWORD2 }}
+      ROBOKASSA_TEST: ${{ vars.PROD_BACKEND_ROBOKASSA_TEST }}
       VITE_VK_APP_ID: ${{ vars.VITE_VK_APP_ID }}
       GATEWAY_VK_ID_CLIENT_SECRET: ${{ secrets.GATEWAY_VK_ID_CLIENT_SECRET }}
       GATEWAY_HONEYTOKEN: ${{ secrets.PROD_GATEWAY_HONEYTOKEN }}

deploy/.env.example

@@ -125,6 +125,17 @@ GATEWAY_VK_APP_SECRET=
 # come from VITE_VK_APP_ID / VITE_VK_ID_REDIRECT_URL above. All three empty disables link.vk.*.
 GATEWAY_VK_ID_CLIENT_SECRET=
 
+# --- Payments: Robokassa (direct RUB rail) ----------------------------------
+# The shop's merchant login + the two pass phrases (Password1 signs the launch request,
+# Password2 signs/verifies the Result callback). An empty login leaves the direct rail off.
+# ROBOKASSA_TEST=1 runs test payments against the shop's TEST pass phrases (no real money);
+# empty/0 is live. Mapped in compose to BACKEND_ROBOKASSA_*; Gitea TEST_/PROD_ secrets, with
+# PROD_BACKEND_ROBOKASSA_TEST a variable so go-live is a flag flip, not a secret redeploy.
+ROBOKASSA_MERCHANT_LOGIN=
+ROBOKASSA_PASSWORD1=
+ROBOKASSA_PASSWORD2=
+ROBOKASSA_TEST=
+
 # --- Gateway anti-abuse ------------------------------------------------------
 # Planted honeytoken bearer value: any request presenting it is flagged — a 24h IP ban
 # where the IP ban is on (prod), logs + a ban metric otherwise (test). Plant the value

deploy/README.md

@@ -76,6 +76,9 @@ compose binds from this directory.
 | `GM_BASICAUTH_HASH` | secret | bcrypt hash gating `/_gm` (admin console + Grafana). Generate with `docker run --rm caddy:2-alpine caddy hash-password --plaintext '<pw>'`. |
 | `TELEGRAM_MINIAPP_URL` | derived | The Mini App URL the bot hands out in deep links / buttons. The deploy derives `PUBLIC_BASE_URL + /telegram/`; set it directly only for a local run (compose still `:?`-requires it). |
 | `EXPORT_SIGN_KEY` | secret | HMAC key signing the public finished-game export download URLs (`/dl/*`). Generate with `openssl rand -base64 32`. |
+| `BACKEND_ROBOKASSA_MERCHANT_LOGIN` | secret | Robokassa shop login for the direct RUB rail. Empty leaves the rail off. |
+| `BACKEND_ROBOKASSA_PASSWORD1` / `…_PASSWORD2` | secret | Robokassa pass phrases: Password1 signs the launch request, Password2 signs/verifies the Result callback. Use the shop's **test** pair while `…_ROBOKASSA_TEST=1`, the **live** pair for real money. (Password3 — Robokassa's JWT-invoice API — is unused.) |
+| `BACKEND_ROBOKASSA_TEST` | **variable** | `1` runs test payments against the test pass phrases (no real money); empty/`0` is live. A variable, not a secret, so go-live is a flag flip + redeploy, not a secret rotation. |
 
 **Plus the bot token** — `TELEGRAM_BOT_TOKEN` (secret), shared by the validator (HMAC
 secret) and the bot (Bot API). It defaults to empty in compose, but both **fail at

deploy/ansible/roles/common/tasks/main.yml

@@ -213,6 +213,20 @@
   loop:
     - ""
     - config
-    - certs
     - dumps
     - images
+
+# The certs dir holds the reverse-mTLS bot-link keypair, bind-mounted into the gateway (and
+# backend) which run as the distroless nonroot UID 65532 — not the deploy user. The dir must be
+# traversable by "other" (0755) or the nonroot process cannot open the 0644 keypair and crash-loops
+# at startup ("mtls: load server keypair: ... permission denied") — a latent failure that only bites
+# on a container restart (e.g. a host reboot), not while a long-running container holds the keypair
+# in memory. The keys themselves are 0644 by design (see the gateway compose); the host is
+# single-tenant and SSH-access-controlled, so a traversable certs dir adds no meaningful exposure.
+- name: Create the scrabble certs directory (traversable by the nonroot gateway UID)
+  ansible.builtin.file:
+    path: "{{ scrabble_base_dir }}/certs"
+    state: directory
+    owner: "{{ deploy_user }}"
+    group: "{{ deploy_user }}"
+    mode: "0755"

deploy/write-prod-env.sh

@@ -52,6 +52,10 @@ export APP_VERSION='$APP_VERSION'
 export TELEGRAM_BOT_TOKEN='$TELEGRAM_BOT_TOKEN'
 export TELEGRAM_MINIAPP_URL='$TELEGRAM_MINIAPP_URL'
 export GATEWAY_VK_APP_SECRET='$GATEWAY_VK_APP_SECRET'
+export ROBOKASSA_MERCHANT_LOGIN='$ROBOKASSA_MERCHANT_LOGIN'
+export ROBOKASSA_PASSWORD1='$ROBOKASSA_PASSWORD1'
+export ROBOKASSA_PASSWORD2='$ROBOKASSA_PASSWORD2'
+export ROBOKASSA_TEST='$ROBOKASSA_TEST'
 export VITE_VK_APP_ID='$VITE_VK_APP_ID'
 export VITE_VK_ID_REDIRECT_URL='$VITE_VK_ID_REDIRECT_URL'
 export GATEWAY_VK_ID_CLIENT_SECRET='$GATEWAY_VK_ID_CLIENT_SECRET'