From 4658e36b1e9df216cb48696a20b3126cfed978ba Mon Sep 17 00:00:00 2001 From: yy-fighting Date: Wed, 8 Jul 2026 15:27:31 +0000 Subject: [PATCH] Fix P800 baseline server startup issues - Fix LOG_DIR unbound variable in long-context matrix start_server.sh. - Add BKCL RDMA env vars to w8a8_int8_baseline mode so cross-device communication initializes correctly (was missing BKCL_RDMA_NICS). --- experiments/dsv4_p800_long_context_matrix/start_server.sh | 4 +++- scripts/common/server_docker.sh | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/experiments/dsv4_p800_long_context_matrix/start_server.sh b/experiments/dsv4_p800_long_context_matrix/start_server.sh index 5b5bcf6..e1bc1bf 100755 --- a/experiments/dsv4_p800_long_context_matrix/start_server.sh +++ b/experiments/dsv4_p800_long_context_matrix/start_server.sh @@ -8,13 +8,15 @@ MAX_RUNNING="${2:-64}" SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" EXPERIMENT_NAME="$(basename "$SCRIPT_DIR")" +RESULT_ROOT="${RESULT_ROOT:-/tmp/${EXPERIMENT_NAME}}" +LOG_DIR="${RESULT_ROOT}/logs" # shellcheck source=/dev/null source "${SCRIPT_DIR}/../../scripts/common/server_docker.sh" # shellcheck source=/dev/null source "${SCRIPT_DIR}/config.env" -log_init "${RESULT_ROOT:-/tmp/${EXPERIMENT_NAME}}/logs/start_server.log" +log_init "${LOG_DIR}/start_server.log" log "starting server for ${EXPERIMENT_NAME}" log "model: ${MODEL_PATH}" log "port: ${PORT}" diff --git a/scripts/common/server_docker.sh b/scripts/common/server_docker.sh index 3d82312..da18480 100755 --- a/scripts/common/server_docker.sh +++ b/scripts/common/server_docker.sh @@ -116,7 +116,7 @@ docker_server_start() { ) # Mode-specific environment variables. - if [[ "$mode" == "w8a8_int8" ]]; then + if [[ "$mode" == "w8a8_int8" || "$mode" == "w8a8_int8_baseline" ]]; then env_args+=( -e SGLANG_DSV4_FP4_EXPERTS=false -e SGLANG_APPLY_CONFIG_BACKUP=auto