fix(deploy): self-healing pre-migrate bootstrap for SecretBackend rollout

Why: clusters upgrading from the pre-SecretBackend schema crash-loop on the
first rollout. `prisma db push` applies the Phase 0 migration as three
sequential steps — add Secret.backendId column (default ''), create
SecretBackend table, add FK — and the FK fails because empty-string values
reference no row in the empty SecretBackend table. This happened on the live
cluster today; I fixed it by hand with psql. This PR makes the fix
automatic so a fresh cluster or anyone replaying the migration doesn't hit
the same trap.

- New `src/db/src/scripts/pre-migrate-bootstrap.ts` — idempotent node script.
  Checks if SecretBackend table exists; if so, ensures a default row exists
  (insert on conflict noop), then backfills any Secret.backendId = '' to
  point at it. Uses Prisma raw queries so it runs against a partially-
  migrated schema.

- `deploy/entrypoint.sh` now catches a failed first push, runs the
  bootstrap, and retries. Fresh installs and fully-migrated clusters take
  the happy path (one push, no bootstrap needed). Pre-Phase-0 upgrades take
  the healing path (push fails → bootstrap seeds → retry succeeds).

- The bootstrap is deliberately non-fatal — even on unexpected errors it
  logs and exits 0 so the retry still runs. If that retry also fails, the
  push error surfaces normally and the pod crash-loops visibly rather than
  silently starting in a half-migrated state.

Verified the idempotent path logically: on the already-bootstrapped cluster
(1 backend row, 0 empty-backendId Secrets), the script's UPDATE matches
zero rows and the INSERT hits ON CONFLICT DO NOTHING — pure no-op.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/db/src/scripts/pre-migrate-bootstrap.ts

@@ -0,0 +1,105 @@
+/**
+ * Self-healing pre-migration step for the SecretBackend rollout (Phase 0).
+ *
+ * Why this exists: `prisma db push` applies schema changes sequentially. When
+ * a cluster upgrades from a pre-SecretBackend DB:
+ *   1. `Secret.backendId` column is added with `DEFAULT ''`
+ *   2. `SecretBackend` table is created (empty)
+ *   3. The FK `Secret.backendId → SecretBackend.id` is added — and FAILS
+ *      because every Secret row now has `backendId = ''` which references no
+ *      row in SecretBackend.
+ *
+ * This script runs AFTER a failed `prisma db push` attempt:
+ *   - If SecretBackend table doesn't exist yet → noop (fresh install case;
+ *     db push will create everything and the FK succeeds because there are
+ *     no Secret rows to violate it).
+ *   - If SecretBackend exists but is empty → insert a default plaintext row.
+ *   - If any Secret rows have `backendId = ''` → point them at the default.
+ *
+ * Idempotent: safe to run multiple times. No-op on a fully-migrated cluster.
+ * Never throws; logs and exits 0 even on errors so the subsequent
+ * `prisma db push` retry is still attempted.
+ */
+import { PrismaClient, Prisma } from '@prisma/client';
+
+const DEFAULT_ID = 'cdefault000backend00000001';
+
+async function main(): Promise<void> {
+  const prisma = new PrismaClient();
+  try {
+    // Does the SecretBackend table exist yet? We check by querying the
+    // information_schema rather than catching Prisma's error — cleaner, and
+    // lets us distinguish "table missing" from "query succeeded but empty".
+    const tableExists = await prisma.$queryRaw<Array<{ exists: boolean }>>`
+      SELECT EXISTS (
+        SELECT 1 FROM information_schema.tables
+        WHERE table_schema = 'public' AND table_name = 'SecretBackend'
+      ) AS exists
+    `;
+    if (!tableExists[0]?.exists) {
+      console.log('bootstrap: SecretBackend table not present yet — skipping');
+      return;
+    }
+
+    // Ensure at least one row exists, marked isDefault.
+    const existingDefault = await prisma.$queryRaw<Array<{ id: string }>>`
+      SELECT id FROM "SecretBackend" WHERE "isDefault" = true LIMIT 1
+    `;
+    let defaultId: string;
+    if (existingDefault.length === 0) {
+      await prisma.$executeRaw`
+        INSERT INTO "SecretBackend"
+          ("id", "name", "type", "config", "isDefault", "description", "version", "createdAt", "updatedAt")
+        VALUES (
+          ${DEFAULT_ID},
+          'default',
+          'plaintext',
+          '{}'::jsonb,
+          true,
+          'Default in-database plaintext backend. Seeded by pre-migrate-bootstrap.',
+          1,
+          CURRENT_TIMESTAMP,
+          CURRENT_TIMESTAMP
+        )
+        ON CONFLICT (name) DO NOTHING
+      `;
+      // Re-read — if there was an existing row with the same name but no
+      // isDefault flag we need its id, not the one we tried to insert.
+      const afterInsert = await prisma.$queryRaw<Array<{ id: string }>>`
+        SELECT id FROM "SecretBackend" WHERE name = 'default' LIMIT 1
+      `;
+      if (afterInsert.length === 0) {
+        console.log('bootstrap: could not establish a default SecretBackend — bailing');
+        return;
+      }
+      defaultId = afterInsert[0]!.id;
+      // Make sure it's flagged default.
+      await prisma.$executeRaw`
+        UPDATE "SecretBackend" SET "isDefault" = true WHERE id = ${defaultId}
+      `;
+      console.log(`bootstrap: seeded default SecretBackend (id=${defaultId})`);
+    } else {
+      defaultId = existingDefault[0]!.id;
+    }
+
+    // Backfill Secret.backendId for any rows left with an empty value.
+    // Using $executeRaw returns affected row count.
+    const updated = await prisma.$executeRaw(
+      Prisma.sql`UPDATE "Secret" SET "backendId" = ${defaultId} WHERE "backendId" = ''`,
+    );
+    if (updated > 0) {
+      console.log(`bootstrap: backfilled ${updated} Secret row(s) with default backendId`);
+    }
+  } catch (err) {
+    // Never fail the deploy — worst case prisma db push tries again anyway.
+    // Log the error so it's visible in pod logs.
+    console.error('bootstrap: non-fatal error:', err instanceof Error ? err.message : err);
+  } finally {
+    await prisma.$disconnect();
+  }
+}
+
+main().catch((err: unknown) => {
+  console.error('bootstrap: fatal error (ignored):', err);
+  // Intentionally exit 0 — we don't want to block the deploy on this.
+});