#!/usr/bin/env bash
#SBATCH --job-name=sass-python-topup
#SBATCH --partition=batch
#SBATCH --exclude=oreus-heimdall-poc-gpu-[004,006]
#SBATCH --nodes=1
#SBATCH --gres=gpu:4
#SBATCH --cpus-per-task=32
#SBATCH --mem=600G
#SBATCH --time=24:00:00
#SBATCH --output=/mnt/vast/maxence/code-search/logs/sass-python-topup-%j.log
#SBATCH --error=/mnt/vast/maxence/code-search/logs/sass-python-topup-%j.log
set -euo pipefail
ARM=${1:?Expected grep or colgrep}
case "$ARM" in grep|colgrep) ;; *) exit 92;; esac
PROJECT=/mnt/vast/maxence/code-search
RUN="$PROJECT/artifacts/generation/python-balanced-20261007"
SHARED_CS=/mnt/vast/maxence/cs-home
cd "$PROJECT"
case "$(hostname -s)" in oreus-heimdall-poc-gpu-003|oreus-heimdall-poc-gpu-005) ;; *) exit 91;; esac
export PATH="$PROJECT/.venv/bin:$SHARED_CS/.cargo/bin:$PATH"
export PYTHONPATH="$RUN/runtime"
export HF_HOME="$SHARED_CS/.cache/huggingface"
export CGD_ARTIFACTS="$PROJECT/artifacts"
export XDG_DATA_HOME="$PROJECT/artifacts/generation/python-multiswe7-rebalanced-15k-deepseek/xdg"
export TMPDIR="/dev/shm/${USER}-sass-python-${SLURM_JOB_ID}"
export CGD_COLGREP_SEARCH_CPU=1 CGD_COLGREP_INDEX_CPU=1 CGD_TOP_K=10
export CGD_MAX_TURNS=10 CGD_COLGREP_CONCURRENCY=8 CGD_COLGREP_THREADS=2
mkdir -p "$TMPDIR"
source "$SHARED_CS/colgrep-gpu/env.sh"
MODEL=deepseek-ai/DeepSeek-V4-Flash-0731
PORT=$((20000 + SLURM_JOB_ID % 10000))
VLOG="$PROJECT/logs/vllm-sass-python-${SLURM_JOB_ID}.log"
python -u "$RUN/generate.py" --preflight-only
VLLM_HOST_IP=127.0.0.1 GLOO_SOCKET_IFNAME=lo NCCL_SOCKET_IFNAME=lo VLLM_ALLREDUCE_USE_SYMM_MEM=0 \
setsid vllm serve "$MODEL" --revision 7872f01b1d1fe23eabc4c98b48bffcef5a386062 \
 --host 127.0.0.1 --port "$PORT" --served-model-name "$MODEL" --trust-remote-code \
 --tensor-parallel-size 4 --kv-cache-dtype fp8 --max-model-len 65536 --max-num-seqs 64 \
 --gpu-memory-utilization 0.9 --disable-custom-all-reduce --enable-prefix-caching \
 --enable-auto-tool-choice --tool-call-parser deepseek_v4 --enforce-eager > "$VLOG" 2>&1 &
SERVER_PID=$!
trap 'kill -- -"$SERVER_PID" 2>/dev/null || true; wait "$SERVER_PID" 2>/dev/null || true' EXIT
ready=0
for ((i=0;i<240;i++)); do
 if curl -sf "http://localhost:$PORT/health" >/dev/null; then ready=1; break; fi
 kill -0 "$SERVER_PID" 2>/dev/null || { tail -60 "$VLOG"; exit 1; }
 sleep 10
done
[[ "$ready" == 1 ]] || { tail -60 "$VLOG"; exit 1; }
python -u "$RUN/generate.py" --base-url "http://localhost:$PORT/v1" --arm "$ARM" --workers 48