EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[CI] Fix PD job (#1129)

Browse Source 
Fix e2e test for Pd job

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-06-09 16:34:41 +08:00
committed by GitHub
parent e63fc6f280
commit cb341c7bcd
3 changed files with 8 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tests/e2e/run_disagg_pd.sh
View File

@@ -43,16 +43,16 @@ _info "Started pd disaggregated proxy server"
PREFILL_PROC_NAME="Prefill-instance"
PREFILL_PORT=8001
run_prefill_instance $MODEL_NAME $TP_SIZE $PREFILL_PORT $REGISTER_PORT $PREFILL_DEVICE_IPS $DECODE_DEVICE_IPS
_info "Starting prefill instance"
run_prefill_instance $MODEL_NAME $TP_SIZE $PREFILL_PORT $REGISTER_PORT $PREFILL_DEVICE_IPS $DECODE_DEVICE_IPS &
_info "Waiting for prefill instance ready"
wait_url_ready $PREFILL_PROC_NAME "http://localhost:${PREFILL_PORT}/v1/completions"
DECODE_PROC_NAME="Decode-instance"
DECODE_PORT=8002
run_decode_instance $MODEL_NAME $TP_SIZE $DECODE_PORT $REGISTER_PORT $PREFILL_DEVICE_IPS $DECODE_DEVICE_IPS
_info "Starting decode instance"
run_decode_instance $MODEL_NAME $TP_SIZE $DECODE_PORT $REGISTER_PORT $PREFILL_DEVICE_IPS $DECODE_DEVICE_IPS &
_info "Waiting for decode instance ready"
wait_url_ready $DECODE_PROC_NAME "http://localhost:${DECODE_PORT}/v1/completions"
_info "pd disaggregated system is ready for handling request"