Portfolio_Risk_Management/deploy/deploy_to_spark.sh
herzogflorian b5da424dfe quantcore: standalone Apptainer setup for students' DGX Sparks
The portfolio data, risk and mean-CVaR optimisation service, prepared for
students who run it on their own DGX Spark without root, Docker or Tailscale.

- deploy/apptainer/qc.sh: install, start/stop/status/logs, cli/api/python,
  data import/export and Yahoo setup/update. cuOpt and quantcore run as two
  Apptainer instances of the user and talk over 127.0.0.1; the API and the
  test bench listen on 127.0.0.1 only. Builds quantcore.sif without root, or
  falls back to a plain Python image plus a virtualenv when the Spark allows
  no unprivileged builds. Detects taken ports and suggests free ones.
- deploy/apptainer/quantcore.def + constraints.txt: the image, with the
  library versions of the verified production container pinned.
- cuOpt under Apptainer: NVIDIA's image sets HOME=/opt/cuopt, read-only in a
  SIF, so the solver died on its first job; qc.sh gives it a writable home.
- solvers: a cuOpt that disappears during a solve now falls back to HiGHS
  instead of answering 500 (requests' ConnectionError escaped the fallback).
- client: idempotent GETs retry a dropped keep-alive connection, which made
  `quantcore-api update --wait` fail when a poll met uvicorn's 5 s timeout.
- config: cuOpt defaults to http://127.0.0.1:5000 and `quantcore serve` to
  127.0.0.1; the Docker images keep passing their own values.
- docs/SPARK_APPTAINER.md: the students' guide; README, USAGE, ARCHITECTURE
  and SPARK_DEPLOYMENT point to it.

Tested on a DGX Spark (GB10, Apptainer 1.5.4, unprivileged user) in both
modes: selftest on every endpoint, optimisation on cuOpt, Yahoo update,
port conflicts; 139 tests pass in the container against the live cuOpt.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-07 09:55:28 +02:00

58 lines
2.1 KiB
Bash
Executable File

#!/usr/bin/env bash
# Deploy the quant service to the DGX Spark:
# 1. rsync the repository to ~/services/work/Portfolio_Risk_Optimization
# 2. optionally rsync an already built data directory (saves the 15-minute setup on the Spark)
# 3. add the `quantcore` service block to ~/services/docker-compose.yml (idempotent)
# 4. build the image and start the container
#
# Usage: deploy/deploy_to_spark.sh [--with-data] [--host user_fh_fhgr@spark-6d4a]
set -euo pipefail
HOST="user_fh_fhgr@spark-6d4a"
WITH_DATA=0
for arg in "$@"; do
case "$arg" in
--with-data) WITH_DATA=1 ;;
--host=*) HOST="${arg#--host=}" ;;
--host) shift; HOST="$1" ;;
esac
done
REPO_DIR="$(cd "$(dirname "$0")/.." && pwd)"
REMOTE_SERVICES="services"
REMOTE_REPO="$REMOTE_SERVICES/work/Portfolio_Risk_Optimization"
REMOTE_DATA="$REMOTE_SERVICES/work/quantcore_data"
echo ">> rsync repository -> $HOST:$REMOTE_REPO"
ssh "$HOST" "mkdir -p $REMOTE_REPO $REMOTE_DATA"
rsync -az --delete \
--exclude '.git' --exclude '/data/' --exclude '__pycache__' --exclude '*.pyc' --exclude '.pytest_cache' \
--exclude '*.egg-info' --exclude '.ruff_cache' --exclude 'scratch/' \
"$REPO_DIR/" "$HOST:$REMOTE_REPO/"
if [[ "$WITH_DATA" == "1" && -d "$REPO_DIR/data" ]]; then
echo ">> rsync data directory -> $HOST:$REMOTE_DATA"
rsync -az --exclude 'jobs/' --exclude '*.log' "$REPO_DIR/data/" "$HOST:$REMOTE_DATA/"
fi
echo ">> register the service in ~/services/docker-compose.yml (if missing)"
ssh "$HOST" bash -s <<'REMOTE'
set -euo pipefail
cd ~/services
if ! grep -q '^ quantcore:' docker-compose.yml; then
cp docker-compose.yml "docker-compose.yml.bak.$(date +%Y%m%d%H%M%S)"
# append the service block (everything below "services:" of the snippet)
awk 'f{print} /^services:/{f=1}' work/Portfolio_Risk_Optimization/deploy/docker-compose.quantcore.yml >> docker-compose.yml
echo " added quantcore service block"
else
echo " quantcore service already present"
fi
docker compose build quantcore
docker compose up -d quantcore
sleep 3
docker compose ps quantcore
curl -s http://localhost:5001/health | head -c 600; echo
REMOTE
echo ">> done. API: http://spark-6d4a:5001/docs"