Skip to content

[Serve] Bugfix flagscale PD disaggregated #1071

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
luoyc123 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[Serve] Bugfix flagscale PD disaggregated #1071

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 28 additions & 6 deletions flagscale/runner/backend/backend_vllm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -784,12 +784,23 @@ def generate_run_script(self, config, host, node_rank, cmd, background=True, wit
p_instance_log_path = os.path.join(default_log_dir, f"prefill_{i}.log")

if update_p_address != master_ip and len(nodes) > 1:
p_kv_config_format_json = p_kv_config_json.replace('"', '\\"')
node_cmd = f"{ids_env} && {vllm_command} --port {http_port} --kv-transfer-config '\\''{p_kv_config_format_json}'\\''"
if docker_name:
p_kv_config_format_json = p_kv_config_json.replace('"', '\\"')
node_cmd = f"{ids_env} && {vllm_command} --port {http_port} --kv-transfer-config '\\''{p_kv_config_format_json}'\\''"
ssh_cmd = f"ssh -f -n -p {ssh_port} {update_p_address} \"docker exec {docker_name} /bin/bash -c '{node_cmd} > {p_instance_log_path} 2>&1 &'\""
else:
ssh_cmd = f'ssh -f -n -p {ssh_port} {update_p_address} "{node_cmd} > {p_instance_log_path} 2>&1 &"'
p_kv_config_format_json = p_kv_config_json.replace(
'"', '\\\\\\"'
)
vllm_command = vllm_command.replace(
"vllm serve", "(vllm serve"
).replace("((vllm serve", "(vllm serve")
node_cmd = f'{vllm_command} --port {http_port} --kv-transfer-config \\"{p_kv_config_format_json}\\" > {p_instance_log_path} 2>&1 &) && disown'
node_cmd = f"{ids_env} && " + node_cmd
logger.info(f"node_cmd {node_cmd}")
ssh_cmd = (
f'ssh -f -n -p {ssh_port} {update_p_address} "{node_cmd}"'
)
f.write(f"{ssh_cmd}\n\n")
else:
p_cmd = f"{ids_env} && {vllm_command} --port {http_port} --kv-transfer-config '\\''{p_kv_config_json}'\\''"
Expand Down Expand Up @@ -843,12 +854,23 @@ def generate_run_script(self, config, host, node_rank, cmd, background=True, wit
d_instance_log_path = os.path.join(default_log_dir, f"decode_{j}.log")

if update_d_address != master_ip and len(nodes) > 1:
d_kv_config_format_json = d_kv_config_json.replace('"', '\\"')
node_cmd = f"{ids_env} && {vllm_command} --port {http_port} --gpu-memory-utilization {decode_gpu_memory_utilization} --kv-transfer-config '\\''{d_kv_config_format_json}'\\''"
if docker_name:
d_kv_config_format_json = d_kv_config_json.replace('"', '\\"')
node_cmd = f"{ids_env} && {vllm_command} --port {http_port} --gpu-memory-utilization {decode_gpu_memory_utilization} --kv-transfer-config '\\''{d_kv_config_format_json}'\\''"
ssh_cmd = f"ssh -f -n -p {ssh_port} {update_d_address} \"docker exec {docker_name} /bin/bash -c '{node_cmd} > {d_instance_log_path} 2>&1 &'\""
else:
ssh_cmd = f'ssh -f -n -p {ssh_port} {update_d_address} "{node_cmd} > {d_instance_log_path} 2>&1 &"'
d_kv_config_format_json = d_kv_config_json.replace(
'"', '\\\\\\"'
)
vllm_command = vllm_command.replace(
"vllm serve", "(vllm serve"
).replace("((vllm serve", "(vllm serve")
node_cmd = f'{vllm_command} --port {http_port} --gpu-memory-utilization {decode_gpu_memory_utilization} --kv-transfer-config \\"{d_kv_config_format_json}\\" > {d_instance_log_path} 2>&1 &) && disown'
node_cmd = f"{ids_env} && " + node_cmd
logger.info(f"node_cmd {node_cmd}")
ssh_cmd = (
f'ssh -f -n -p {ssh_port} {update_d_address} "{node_cmd}"'
)
f.write(f"{ssh_cmd}\n\n")
else:
d_cmd = f"{ids_env} && {vllm_command} --port {http_port} --gpu-memory-utilization {decode_gpu_memory_utilization} --kv-transfer-config '\\''{d_kv_config_json}'\\''"
Expand Down