Fix migration failures: switch to full TimescaleDB + use psql runner

- Change image from timescaledb-ha:pg16-ts2.15-oss to pg16-ts2.15
  (OSS edition lacks compression, retention, continuous aggregates)
- Add postgresql-client to Dockerfile for psql binary
- Rewrite run_migrations.py to use psql instead of psycopg2
  (psql runs each statement independently; psycopg2 wraps the
  entire file in one transaction so one error rolls back everything)
- Add schema verification: exits 1 if critical tables missing,
  preventing services from starting with broken schema

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Dockerfile

@@ -7,6 +7,7 @@ COPY --from=ghcr.io/astral-sh/uv:latest /uv /bin/uv
 # Install system dependencies (Required for Postgres and Healthchecks)
 RUN apt-get update && apt-get install -y \
     libpq5 \
+    postgresql-client \
     curl \
     && rm -rf /var/lib/apt/lists/*
 

docker-compose.yaml

@@ -1,6 +1,6 @@
 services:
   timescale_db:
-    image: timescale/timescaledb-ha:pg16-ts2.15-oss
+    image: timescale/timescaledb-ha:pg16-ts2.15
     restart: always
     environment:
       - POSTGRES_DB=${POSTGRES_DB}

run_migrations.py

@@ -1,11 +1,13 @@
 """
-run_migrations.py — Idempotent SQL migration runner for Docker init service.
-Executes each .sql migration file in order using psycopg2.
-Tolerates re-run errors (e.g. "policy already exists") so deploys are safe.
+run_migrations.py — Idempotent SQL migration runner for Docker init.
+Uses psql (not psycopg2) so each statement runs independently —
+one error doesn't roll back the entire file.
 """
 
 import os
+import subprocess
 import sys
+
 import psycopg2
 
 DATABASE_URL = os.environ["DATABASE_URL"]
@@ -15,51 +17,71 @@ MIGRATIONS = [
     "03_webhook_schema_migration.sql",
 ]
 
+CRITICAL_TABLES = [
+    "tracksolid.devices",
+    "tracksolid.api_token_cache",
+    "tracksolid.ingestion_log",
+    "tracksolid.live_positions",
+    "tracksolid.position_history",
+    "tracksolid.trips",
+    "tracksolid.alarms",
+    "tracksolid.obd_readings",
+]
 
-def run_file(conn, path, filename):
-    """Execute a SQL file. Returns True on success, False on error."""
-    with open(path) as f:
-        sql = f.read()
-    try:
-        with conn.cursor() as cur:
-            cur.execute(sql)
-        print(f"  OK: {filename}")
-        return True
-    except psycopg2.Error as e:
-        msg = (e.pgerror or str(e)).strip().split("\n")[0]
-        print(f"  WARN: {filename}: {msg}")
-        # Connection is now in error state — must reset
-        conn.close()
+
+def run_file(path, filename):
+    """Execute a SQL file via psql. Returns True on success."""
+    print(f"Running {filename}...")
+    result = subprocess.run(
+        ["psql", DATABASE_URL, "-f", path],
+        capture_output=True, text=True,
+    )
+    # psql prints errors to stderr but continues by default
+    errors = [l for l in result.stderr.splitlines() if "ERROR:" in l]
+    if errors:
+        for e in errors:
+            print(f"  WARN: {e.strip()}")
         return False
+    print(f"  OK: {filename}")
+    return True
+
+
+def verify_schema():
+    """Verify critical tables exist. Exit 1 if not — prevents services from starting."""
+    print("Verifying schema...")
+    conn = psycopg2.connect(DATABASE_URL)
+    cur = conn.cursor()
+    missing = []
+    for table in CRITICAL_TABLES:
+        schema, name = table.split(".")
+        cur.execute(
+            "SELECT 1 FROM information_schema.tables WHERE table_schema=%s AND table_name=%s",
+            (schema, name),
+        )
+        if not cur.fetchone():
+            missing.append(table)
+    cur.close()
+    conn.close()
+
+    if missing:
+        print(f"FATAL: Missing critical tables: {', '.join(missing)}")
+        print("Schema bootstrap failed. Services cannot start.")
+        sys.exit(1)
+    print("  All critical tables verified.")
 
 
 def main():
     print("=== Database Migration Runner ===")
-    conn = psycopg2.connect(DATABASE_URL)
-    conn.autocommit = True
 
-    warnings = 0
     for sql_file in MIGRATIONS:
         path = os.path.join("/app", sql_file)
         if not os.path.exists(path):
             print(f"  SKIP: {sql_file} (not found)")
             continue
+        run_file(path, sql_file)
 
-        print(f"Running {sql_file}...")
-        ok = run_file(conn, path, sql_file)
-        if not ok:
-            warnings += 1
-            # Reconnect for the next file
-            conn = psycopg2.connect(DATABASE_URL)
-            conn.autocommit = True
-
-    conn.close()
-
-    if warnings:
-        print(f"Completed with {warnings} warning(s) (expected on re-deploy).")
-    else:
-        print("All migrations applied cleanly.")
-    sys.exit(0)
+    verify_schema()
+    print("Migrations complete.")
 
 
 if __name__ == "__main__":