fix(ci): align deploy-dev/deploy-uat frontend bump with base image name (CAR-1318)

The base frontend deployment in cartsnitch/infra uses
ghcr.io/cartsnitch/app, but the deploy-dev/deploy-uat jobs
targeted ghcr.io/cartsnitch/cartsnitch — a name that matches no
resource, so each bump added a dead images entry and left the
real 'app' entry's newTag pinned.

Switch both jobs to kustomize edit set image
ghcr.io/cartsnitch/app=... so the next deploy updates the active
entry. The matching infra overlay cleanup drops the dead
cartsnitch entry from dev/uat.

Co-Authored-By: Paperclip <noreply@paperclip.ing>

.gitea/workflows/ci.yml

@@ -26,11 +26,7 @@ jobs:
   lint:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-node@v4
-        with:
-          node-version: "20"
-          cache: npm
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
       - run: npm ci
       - name: ESLint
         run: npx eslint .
@@ -40,8 +36,8 @@ jobs:
   test:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-node@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
+      - uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
         with:
           node-version: "20"
           cache: npm
@@ -52,8 +48,8 @@ jobs:
   audit:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-node@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
+      - uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
         with:
           node-version: "20"
           cache: npm
@@ -64,8 +60,8 @@ jobs:
   e2e:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-node@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
+      - uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
         with:
           node-version: "20"
           cache: npm
@@ -77,8 +73,8 @@ jobs:
     runs-on: ubuntu-latest
     needs: [test]
     steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-node@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
+      - uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020
         with:
           node-version: "20"
           cache: npm
@@ -106,7 +102,7 @@ jobs:
       calver_tag: ${{ steps.calver.outputs.version }}
       sha_tag: sha-${{ github.sha }}
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
         with:
           fetch-depth: 0
 
@@ -160,8 +156,8 @@ jobs:
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
           target: prod
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
+          cache-from: type=inline
+          cache-to: type=inline,mode=max
 
       - name: Scan frontend image for vulnerabilities
         uses: anchore/scan-action@v5
@@ -186,7 +182,7 @@ jobs:
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
           target: prod
-          cache-from: type=gha
+          cache-from: type=inline
 
       - name: Create git tag
         if: github.event_name == 'push' && github.ref == 'refs/heads/main'
@@ -202,7 +198,7 @@ jobs:
       calver_tag: ${{ steps.calver.outputs.version }}
       sha_tag: sha-${{ github.sha }}
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
         with:
           fetch-depth: 0
 
@@ -252,8 +248,8 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
             APT_CACHE_BUST=${{ github.run_id }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
+          cache-from: type=inline
+          cache-to: type=inline,mode=max
 
       - name: Scan receiptwitness image for vulnerabilities
         uses: anchore/scan-action@v5
@@ -280,7 +276,7 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
             APT_CACHE_BUST=${{ github.run_id }}
-          cache-from: type=gha
+          cache-from: type=inline
 
   build-and-push-api:
     runs-on: ubuntu-latest
@@ -290,7 +286,7 @@ jobs:
       calver_tag: ${{ steps.calver.outputs.version }}
       sha_tag: sha-${{ github.sha }}
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
         with:
           fetch-depth: 0
 
@@ -340,8 +336,8 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
             APT_CACHE_BUST=${{ github.run_id }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
+          cache-from: type=inline
+          cache-to: type=inline,mode=max
 
       - name: Scan api image for vulnerabilities
         uses: anchore/scan-action@v5
@@ -368,7 +364,7 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
             APT_CACHE_BUST=${{ github.run_id }}
-          cache-from: type=gha
+          cache-from: type=inline
 
   build-and-push-auth:
     runs-on: ubuntu-latest
@@ -378,7 +374,7 @@ jobs:
       calver_tag: ${{ steps.calver.outputs.version }}
       sha_tag: sha-${{ github.sha }}
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
         with:
           fetch-depth: 0
 
@@ -428,8 +424,8 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
             APT_CACHE_BUST=${{ github.run_id }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
+          cache-from: type=inline
+          cache-to: type=inline,mode=max
 
       - name: Scan auth image for vulnerabilities
         uses: anchore/scan-action@v5
@@ -456,7 +452,7 @@ jobs:
           labels: ${{ steps.meta.outputs.labels }}
           build-args: |
             APT_CACHE_BUST=${{ github.run_id }}
-          cache-from: type=gha
+          cache-from: type=inline
 
   deploy-dev:
     runs-on: ubuntu-latest
@@ -464,10 +460,10 @@ jobs:
     if: always() && !cancelled() && github.event_name == 'push' && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main')
     steps:
       - name: Checkout infra repo
-        uses: actions/checkout@v4
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
         with:
           repository: cartsnitch/infra
-          token: ${{ secrets.REGISTRY_TOKEN }}
+          token: ${{ secrets.CI_GITEA_TOKEN }}
           ref: main
           path: infra
 
@@ -475,7 +471,16 @@ jobs:
         uses: azure/setup-kubectl@v4
 
       - name: Install kustomize
-        uses: imranismail/setup-kustomize@v2
+        # imranismail/setup-kustomize@v2 calls the Gitea API to record
+        # telemetry under the "kubernetes-sigs" user, which doesn't exist
+        # on this Gitea instance. Install the binary directly instead.
+        run: |
+          set -euo pipefail
+          version="5.4.3"
+          url="https://github.com/kubernetes-sigs/kustomize/releases/download/kustomize%2Fv${version}/kustomize_v${version}_linux_amd64.tar.gz"
+          curl -fsSL --retry 3 "$url" | tar -xz -C /tmp kustomize
+          sudo install -m 0755 /tmp/kustomize /usr/local/bin/kustomize
+          kustomize version
 
       - name: Determine image tag for frontend
         id: frontend_tag
@@ -490,7 +495,7 @@ jobs:
         if: needs.build-and-push.result == 'success'
         run: |
           cd infra/apps/overlays/dev
-          kustomize edit set image ghcr.io/cartsnitch/cartsnitch=git.farh.net/cartsnitch/cartsnitch:${{ steps.frontend_tag.outputs.tag }}
+          kustomize edit set image ghcr.io/cartsnitch/app=git.farh.net/cartsnitch/cartsnitch:${{ steps.frontend_tag.outputs.tag }}
 
       - name: Determine image tag for receiptwitness
         id: receiptwitness_tag
@@ -537,16 +542,63 @@ jobs:
           cd infra/apps/overlays/dev
           kustomize edit set image ghcr.io/cartsnitch/auth=git.farh.net/cartsnitch/auth:${{ steps.auth_tag.outputs.tag }}
 
-      - name: Commit and push to infra
+      - name: Commit and push to infra (via PR)
+        env:
+          CI_GITEA_TOKEN: ${{ secrets.CI_GITEA_TOKEN }}
         run: |
           cd infra
           git config user.name "cartsnitch-ci[bot]"
           git config user.email "cartsnitch-ci[bot]@users.noreply.git.farh.net"
           git add apps/overlays/dev/kustomization.yaml
           git diff --cached --quiet && echo "No image changes to deploy" && exit 0
+          BRANCH="ci/deploy-dev-${GITHUB_SHA}"
+          git checkout -b "$BRANCH"
           git commit -m "ci(dev): update cartsnitch, receiptwitness, api, and auth images"
-          git pull --rebase origin main
-          git push origin main
+          git push origin "$BRANCH"
+          PR_BODY=$(printf 'Auto-opened by deploy-dev (CAR-1195).\n\nBuild SHA: %s' "${GITHUB_SHA}")
+          PR_JSON=$(curl -sS -X POST \
+            -H "Authorization: token ${CI_GITEA_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d "$(jq -n --arg head "cartsnitch:${BRANCH}" --arg base main --arg title "ci(dev): update overlay image tags (${GITHUB_SHA::12})" --arg body "$PR_BODY" '{head:$head,base:$base,title:$title,body:$body}')" \
+            "https://git.farh.net/api/v1/repos/cartsnitch/infra/pulls")
+          PR_NUM=$(echo "$PR_JSON" | jq -r '.number // empty')
+          if [ -z "$PR_NUM" ]; then
+            echo "::error::Failed to open PR against cartsnitch/infra: $PR_JSON"
+            exit 1
+          fi
+          echo "Opened cartsnitch/infra PR #${PR_NUM} (head=${BRANCH})"
+          # Request CTO (cs_savannah) review as the GitOps hand-off. Best-effort:
+          # log on non-2xx but never fail the job for this.
+          REVIEW_HTTP=$(curl -sS -o /dev/null -w '%{http_code}' -X POST \
+            -H "Authorization: token ${CI_GITEA_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d '{"reviewers":["cs_savannah"]}' \
+            "https://git.farh.net/api/v1/repos/cartsnitch/infra/pulls/${PR_NUM}/requested_reviewers")
+          if [ "${REVIEW_HTTP}" -lt 200 ] || [ "${REVIEW_HTTP}" -ge 300 ]; then
+            echo "::notice::Failed to request reviewers for cartsnitch/infra PR #${PR_NUM} (HTTP ${REVIEW_HTTP}); continuing"
+          fi
+          # CAR-1216: the in-job merge attempt is a best-effort fast-path only.
+          # `cartsnitch/infra` main requires a human approving review (immutable
+          # branch protection); the CI bot (`CI_GITEA_TOKEN`) can never self-
+          # approve, so this merge call structurally cannot succeed in the
+          # general case. Any non-merged outcome (approvals pending, checks
+          # pending, any other Gitea message) is the GitOps approval gate, not
+          # a CI failure — the PR is already opened and `cs_savannah` is
+          # requested as reviewer above. Surface the response as a notice and
+          # exit success. The only hard-fail (`exit 1`) in this step remains
+          # the empty-`PR_NUM` check (PR could not be created at all).
+          MERGE_RESP=$(curl -sS -X POST \
+            -H "Authorization: token ${CI_GITEA_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d '{"Do":"merge","delete_branch_after_merge":true}' \
+            "https://git.farh.net/api/v1/repos/cartsnitch/infra/pulls/${PR_NUM}/merge")
+          MERGED=$(echo "$MERGE_RESP" | jq -r '.merged // false')
+          if [ "$MERGED" = "true" ]; then
+            echo "PR #${PR_NUM} merged into cartsnitch/infra main"
+          else
+            echo "::notice::infra PR #${PR_NUM} opened and awaiting CTO (cs_savannah) approve+merge — GitOps approval gate, not a failure: $MERGE_RESP"
+            exit 0
+          fi
 
   deploy-uat:
     runs-on: ubuntu-latest
@@ -554,10 +606,10 @@ jobs:
     if: always() && !cancelled() && github.event_name == 'push' && (github.ref == 'refs/heads/uat' || github.ref == 'refs/heads/main')
     steps:
       - name: Checkout infra repo
-        uses: actions/checkout@v4
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
         with:
           repository: cartsnitch/infra
-          token: ${{ secrets.REGISTRY_TOKEN }}
+          token: ${{ secrets.CI_GITEA_TOKEN }}
           ref: main
           path: infra
 
@@ -565,7 +617,16 @@ jobs:
         uses: azure/setup-kubectl@v4
 
       - name: Install kustomize
-        uses: imranismail/setup-kustomize@v2
+        # imranismail/setup-kustomize@v2 calls the Gitea API to record
+        # telemetry under the "kubernetes-sigs" user, which doesn't exist
+        # on this Gitea instance. Install the binary directly instead.
+        run: |
+          set -euo pipefail
+          version="5.4.3"
+          url="https://github.com/kubernetes-sigs/kustomize/releases/download/kustomize%2Fv${version}/kustomize_v${version}_linux_amd64.tar.gz"
+          curl -fsSL --retry 3 "$url" | tar -xz -C /tmp kustomize
+          sudo install -m 0755 /tmp/kustomize /usr/local/bin/kustomize
+          kustomize version
 
       - name: Determine image tag for frontend
         id: frontend_tag
@@ -580,7 +641,7 @@ jobs:
         if: needs.build-and-push.result == 'success'
         run: |
           cd infra/apps/overlays/uat
-          kustomize edit set image ghcr.io/cartsnitch/cartsnitch=git.farh.net/cartsnitch/cartsnitch:${{ steps.frontend_tag.outputs.tag }}
+          kustomize edit set image ghcr.io/cartsnitch/app=git.farh.net/cartsnitch/cartsnitch:${{ steps.frontend_tag.outputs.tag }}
 
       - name: Determine image tag for receiptwitness
         id: receiptwitness_tag
@@ -627,13 +688,60 @@ jobs:
           cd infra/apps/overlays/uat
           kustomize edit set image ghcr.io/cartsnitch/auth=git.farh.net/cartsnitch/auth:${{ steps.auth_tag.outputs.tag }}
 
-      - name: Commit and push to infra
+      - name: Commit and push to infra (via PR)
+        env:
+          CI_GITEA_TOKEN: ${{ secrets.CI_GITEA_TOKEN }}
         run: |
           cd infra
           git config user.name "cartsnitch-ci[bot]"
           git config user.email "cartsnitch-ci[bot]@users.noreply.git.farh.net"
           git add apps/overlays/uat/kustomization.yaml
           git diff --cached --quiet && echo "No image changes to deploy" && exit 0
+          BRANCH="ci/deploy-uat-${GITHUB_SHA}"
+          git checkout -b "$BRANCH"
           git commit -m "ci(uat): update cartsnitch, receiptwitness, api, and auth images"
-          git pull --rebase origin main
-          git push origin main
+          git push origin "$BRANCH"
+          PR_BODY=$(printf 'Auto-opened by deploy-uat (CAR-1195).\n\nBuild SHA: %s' "${GITHUB_SHA}")
+          PR_JSON=$(curl -sS -X POST \
+            -H "Authorization: token ${CI_GITEA_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d "$(jq -n --arg head "cartsnitch:${BRANCH}" --arg base main --arg title "ci(uat): update overlay image tags (${GITHUB_SHA::12})" --arg body "$PR_BODY" '{head:$head,base:$base,title:$title,body:$body}')" \
+            "https://git.farh.net/api/v1/repos/cartsnitch/infra/pulls")
+          PR_NUM=$(echo "$PR_JSON" | jq -r '.number // empty')
+          if [ -z "$PR_NUM" ]; then
+            echo "::error::Failed to open PR against cartsnitch/infra: $PR_JSON"
+            exit 1
+          fi
+          echo "Opened cartsnitch/infra PR #${PR_NUM} (head=${BRANCH})"
+          # Request CTO (cs_savannah) review as the GitOps hand-off. Best-effort:
+          # log on non-2xx but never fail the job for this.
+          REVIEW_HTTP=$(curl -sS -o /dev/null -w '%{http_code}' -X POST \
+            -H "Authorization: token ${CI_GITEA_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d '{"reviewers":["cs_savannah"]}' \
+            "https://git.farh.net/api/v1/repos/cartsnitch/infra/pulls/${PR_NUM}/requested_reviewers")
+          if [ "${REVIEW_HTTP}" -lt 200 ] || [ "${REVIEW_HTTP}" -ge 300 ]; then
+            echo "::notice::Failed to request reviewers for cartsnitch/infra PR #${PR_NUM} (HTTP ${REVIEW_HTTP}); continuing"
+          fi
+          # CAR-1216: the in-job merge attempt is a best-effort fast-path only.
+          # `cartsnitch/infra` main requires a human approving review (immutable
+          # branch protection); the CI bot (`CI_GITEA_TOKEN`) can never self-
+          # approve, so this merge call structurally cannot succeed in the
+          # general case. Any non-merged outcome (approvals pending, checks
+          # pending, any other Gitea message) is the GitOps approval gate, not
+          # a CI failure — the PR is already opened and `cs_savannah` is
+          # requested as reviewer above. Surface the response as a notice and
+          # exit success. The only hard-fail (`exit 1`) in this step remains
+          # the empty-`PR_NUM` check (PR could not be created at all).
+          MERGE_RESP=$(curl -sS -X POST \
+            -H "Authorization: token ${CI_GITEA_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d '{"Do":"merge","delete_branch_after_merge":true}' \
+            "https://git.farh.net/api/v1/repos/cartsnitch/infra/pulls/${PR_NUM}/merge")
+          MERGED=$(echo "$MERGE_RESP" | jq -r '.merged // false')
+          if [ "$MERGED" = "true" ]; then
+            echo "PR #${PR_NUM} merged into cartsnitch/infra main"
+          else
+            echo "::notice::infra PR #${PR_NUM} opened and awaiting CTO (cs_savannah) approve+merge — GitOps approval gate, not a failure: $MERGE_RESP"
+            exit 0
+          fi

auth/Dockerfile

@@ -1,4 +1,4 @@
-FROM node:22-alpine AS builder
+FROM node:22-alpine@sha256:8ea2348b068a9544dae7317b4f3aafcdc032df1647bb7d768a05a5cad1a7683f AS builder
 RUN apk update && apk upgrade --no-cache
 WORKDIR /app
 COPY package.json package-lock.json* ./
@@ -7,7 +7,7 @@ COPY tsconfig.json ./
 COPY src/ src/
 RUN npm run build
 
-FROM node:22-alpine
+FROM node:22-alpine@sha256:8ea2348b068a9544dae7317b4f3aafcdc032df1647bb7d768a05a5cad1a7683f
 RUN apk update && apk upgrade --no-cache
 WORKDIR /app
 ENV NODE_ENV=production

auth/src/__tests__/health.test.ts

@@ -19,9 +19,18 @@ describe('Auth health endpoint', () => {
             }
             res.writeHead(200, { 'Content-Type': 'application/json' });
             res.end(JSON.stringify({ status: 'ok', db: 'reachable' }));
-          } catch {
+          } catch (err) {
+            // Mirror src/index.ts: log the error and include the message in the
+            // response body so /health 503s are diagnosable from pod logs.
+            console.error(
+              '[auth /health] DB probe failed:',
+              err instanceof Error ? `${err.name}: ${err.message}` : err,
+            );
+            const detail = err instanceof Error ? err.message : 'unknown error';
             res.writeHead(503, { 'Content-Type': 'application/json' });
-            res.end(JSON.stringify({ status: 'error', db: 'unreachable' }));
+            res.end(
+              JSON.stringify({ status: 'error', db: 'unreachable', error: detail }),
+            );
           }
           return;
         }
@@ -76,7 +85,10 @@ describe('Auth health endpoint', () => {
     close();
 
     equal(status, 503);
-    equal(body, '{"status":"error","db":"unreachable"}');
+    const parsed = JSON.parse(body);
+    equal(parsed.status, 'error');
+    equal(parsed.db, 'unreachable');
+    equal(parsed.error, 'connection refused');
   });
 
   it('returns 503 with db=unreachable when query times out', async () => {
@@ -95,7 +107,14 @@ describe('Auth health endpoint', () => {
     close();
 
     equal(status, 503);
-    equal(body, '{"status":"error","db":"unreachable"}');
+    const parsed = JSON.parse(body);
+    equal(parsed.status, 'error');
+    equal(parsed.db, 'unreachable');
+    // The query promise rejects with a synthetic 'timeout' error; the
+    // Promise.race wrapper also rejects with 'DB timeout'. The body should
+    // surface whichever error was thrown — accept either to stay robust.
+    equal(typeof parsed.error, 'string');
+    equal(parsed.error.length > 0, true);
   });
 
   it('returns a terminal response for unknown paths (no hang)', async () => {

auth/src/index.ts

@@ -21,9 +21,19 @@ const server = createServer(async (req, res) => {
       }
       res.writeHead(200, { "Content-Type": "application/json" });
       res.end(JSON.stringify({ status: "ok", db: "reachable" }));
-    } catch {
+    } catch (err) {
+      // Log the actual error so /health 503s are diagnosable from pod logs
+      // (CAR-1276: UAT auth was crashlooping with no log output beyond the
+      // initial "listening on port 3001" line because this catch was empty).
+      console.error(
+        "[auth /health] DB probe failed:",
+        err instanceof Error ? `${err.name}: ${err.message}` : err,
+      );
+      const detail = err instanceof Error ? err.message : "unknown error";
       res.writeHead(503, { "Content-Type": "application/json" });
-      res.end(JSON.stringify({ status: "error", db: "unreachable" }));
+      res.end(
+        JSON.stringify({ status: "error", db: "unreachable", error: detail }),
+      );
     }
     return;
   }

common/src/cartsnitch_common/database.py

@@ -1,17 +1,36 @@
 """Database engine and session factories for sync and async usage."""
 
 from collections.abc import AsyncGenerator, Generator
+from typing import TYPE_CHECKING
 
 from sqlalchemy import create_engine
-from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
+from sqlalchemy.ext.asyncio import AsyncEngine, AsyncSession, async_sessionmaker, create_async_engine
 from sqlalchemy.orm import Session, sessionmaker
 
 from cartsnitch_common.config import settings
 
+if TYPE_CHECKING:
+    from sqlalchemy.engine import Engine
 
-def get_async_engine(url: str | None = None):
-    """Create an async SQLAlchemy engine."""
-    return create_async_engine(url or settings.database_url, echo=settings.debug)
+# Module-level async engine cache — one engine per unique URL, shared across all callers.
+# This prevents pool exhaustion in high-throughput workers (e.g. email-worker hitting
+# DragonflyDB/Postgres repeatedly per message).  pool_size=10, max_overflow=20 gives
+# headroom for bursts while capping max connections at 30 per URL.
+_async_engine_cache: dict[str, "AsyncEngine"] = {}
+
+
+def get_async_engine(url: str | None = None) -> "AsyncEngine":
+    """Get or create a cached async engine for the given URL."""
+    target = url or settings.database_url
+    if target not in _async_engine_cache:
+        _async_engine_cache[target] = create_async_engine(
+            target,
+            echo=settings.debug,
+            pool_size=10,
+            max_overflow=20,
+            pool_pre_ping=True,
+        )
+    return _async_engine_cache[target]
 
 
 def get_sync_engine(url: str | None = None):

package.json

@@ -45,14 +45,16 @@
     "typescript-eslint": "^8.56.1",
     "vite": "^6.4.2",
     "vite-plugin-pwa": "^0.21.2",
-    "vitest": "^3.2.4"
+    "vitest": "^4.1.8"
   },
   "overrides": {
     "@rollup/pluginutils": "5.3.0",
     "flatted": "^3.4.2",
     "serialize-javascript": "7.0.5",
-    "brace-expansion": ">=1.1.13",
+    "brace-expansion": ">=5.0.6",
     "lodash": ">=4.17.24",
-    "minimatch": "^10.2.4"
+    "minimatch": "^10.2.4",
+    "@babel/plugin-transform-modules-systemjs": "^7.29.4",
+    "fast-uri": "^3.1.2"
   }
 }

receiptwitness/src/receiptwitness/queue/email.py

@@ -16,6 +16,29 @@ logger = logging.getLogger(__name__)
 STREAM_KEY = "email:receipts"
 CONSUMER_GROUP = "email-workers"
 
+# Module-level Redis/DragonflyDB connection pool — shared across all worker calls.
+# Without pooling, each call to get_redis() opens a new TCP connection.  In a tight
+# consumer loop this causes ConnectionResetError when DragonflyDB's connection limit
+# is hit under load.  max_connections=30 (10 base + 20 overflow) mirrors the engine pool.
+_redis_pool: aioredis.ConnectionPool | None = None
+
+
+def _get_redis_pool() -> aioredis.ConnectionPool:
+    """Get or create the shared DragonflyDB connection pool."""
+    global _redis_pool
+    if _redis_pool is None:
+        _redis_pool = aioredis.ConnectionPool.from_url(
+            settings.redis_url,
+            decode_responses=True,
+            max_connections=30,
+        )
+    return _redis_pool
+
+
+async def get_redis() -> aioredis.Redis:
+    """Get async Redis/DragonflyDB client backed by a shared connection pool."""
+    return aioredis.Redis(connection_pool=_get_redis_pool())
+
 
 @dataclass
 class EmailJob:
@@ -31,11 +54,6 @@ class EmailJob:
     message_id: str  # from email provider, for dedup
 
 
-async def get_redis() -> aioredis.Redis:
-    """Get async Redis/DragonflyDB client."""
-    return cast(aioredis.Redis, aioredis.from_url(settings.redis_url, decode_responses=True))
-
-
 async def ensure_consumer_group(client: aioredis.Redis) -> None:
     """Create consumer group if it does not exist."""
     try: