Files
beanflows/infra/supervisor/supervisor.sh
Deeman 9ee7a3d9d3 fix: export_serving — Arrow-based copy, rename to analytics.duckdb
Two bugs fixed:

1. Cross-connection COPY: DuckDB doesn't support referencing another
   connection's tables as src.serving.table. Replace with Arrow as
   intermediate: src reads to Arrow, dst.register() + CREATE TABLE.

2. Catalog/schema name collision: naming the export file serving.duckdb
   made DuckDB assign catalog name "serving" — same as the schema we
   create inside it. Every serving.table query became ambiguous. Rename
   to analytics.duckdb (catalog "analytics", schema "serving" = no clash).

   SERVING_DUCKDB_PATH values updated: serving.duckdb → analytics.duckdb
   in supervisor, service, bootstrap, dev_run.sh, .env.example, docker-compose.

3. Temp file: use _export.duckdb (not serving.duckdb.tmp) to avoid
   the same catalog collision during the write phase.

Verified: 6 tables exported, serving.* queries work read-only.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-22 12:54:39 +01:00

70 lines
2.7 KiB
Bash

#!/bin/sh
# Materia Supervisor - Continuous pipeline orchestration
# Inspired by TigerBeetle's CFO supervisor: simple, resilient, easy to understand
# https://github.com/tigerbeetle/tigerbeetle/blob/main/src/scripts/cfo_supervisor.sh
#
# Environment variables (set in systemd EnvironmentFile):
# LANDING_DIR — local path for extracted landing data
# DUCKDB_PATH — path to DuckDB lakehouse file (SQLMesh pipeline DB)
# SERVING_DUCKDB_PATH — path to serving-only DuckDB (web app reads from here)
# ALERT_WEBHOOK_URL — optional ntfy.sh / Slack / Telegram webhook for failure alerts
set -eu
readonly REPO_DIR="/opt/materia"
while true
do
(
# Clone repo if missing
if ! [ -d "$REPO_DIR/.git" ]
then
echo "Repository not found, bootstrap required!"
exit 1
fi
cd "$REPO_DIR"
# Update code from git
git fetch origin master
git switch --discard-changes --detach origin/master
uv sync
# Extract all data sources
LANDING_DIR="${LANDING_DIR:-/data/materia/landing}" \
DUCKDB_PATH="${DUCKDB_PATH:-/data/materia/lakehouse.duckdb}" \
uv run materia pipeline run extract
LANDING_DIR="${LANDING_DIR:-/data/materia/landing}" \
DUCKDB_PATH="${DUCKDB_PATH:-/data/materia/lakehouse.duckdb}" \
uv run materia pipeline run extract_cot
LANDING_DIR="${LANDING_DIR:-/data/materia/landing}" \
DUCKDB_PATH="${DUCKDB_PATH:-/data/materia/lakehouse.duckdb}" \
uv run materia pipeline run extract_prices
LANDING_DIR="${LANDING_DIR:-/data/materia/landing}" \
DUCKDB_PATH="${DUCKDB_PATH:-/data/materia/lakehouse.duckdb}" \
uv run materia pipeline run extract_ice
# Transform all data sources
LANDING_DIR="${LANDING_DIR:-/data/materia/landing}" \
DUCKDB_PATH="${DUCKDB_PATH:-/data/materia/lakehouse.duckdb}" \
uv run materia pipeline run transform
# Export serving tables to analytics.duckdb (atomic swap).
# The web app reads from SERVING_DUCKDB_PATH and picks up the new file
# automatically via inode-based connection reopen — no restart needed.
DUCKDB_PATH="${DUCKDB_PATH:-/data/materia/lakehouse.duckdb}" \
SERVING_DUCKDB_PATH="${SERVING_DUCKDB_PATH:-/data/materia/analytics.duckdb}" \
uv run materia pipeline run export_serving
) || {
# Notify on failure if webhook is configured, then sleep to avoid busy-loop
if [ -n "${ALERT_WEBHOOK_URL:-}" ]; then
curl -s -d "Materia pipeline failed at $(date)" "$ALERT_WEBHOOK_URL" 2>/dev/null || true
fi
sleep 600 # Sleep 10 min on failure
}
done