upstream-archive/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-29 17:37:39 -05:00
Code Issues Packages Projects Releases Wiki Activity

server: add --sleep-mode rst

Browse Source
This commit is contained in:
Xuan Son Nguyen
2026-08-20 00:37:08 +02:00
parent 5f9b61d603
commit 437b415fa8
10 changed files with 350 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/server/server.cpp
+5 -4
View File
@@ -167,6 +167,7 @@ int llama_server(common_params & params, int argc, char ** argv) {
// struct that contains llama context and inference
server_context ctx_server;
ctx_server.init(argc, argv);
server_http_context ctx_http;
if (!ctx_http.init(params)) {
@@ -458,11 +459,8 @@ int llama_server(common_params & params, int argc, char ** argv) {
return 1;
}
routes.update_meta(ctx_server);
ctx_http.is_ready.store(true);
SRV_INF("%s", "model loaded\n");
shutdown_handler = [&](int) {
mcp_mgr.shutdown();
// this will unblock start_loop()
@@ -513,7 +511,10 @@ int llama_server(common_params & params, int argc, char ** argv) {
std::thread monitor_thread;
if (child.is_child()) {
monitor_thread = child.setup(shutdown_handler);
child.notify_to_router(server_state_to_str(SERVER_STATE_READY), routes.get_model_info());
// if no model is loaded, the process restarted into sleeping state and the router knows it
if (ctx_server.get_llama_context() != nullptr) {
child.notify_to_router(server_state_to_str(SERVER_STATE_READY), routes.get_model_info());
}
}
// this call blocks the main thread until queue_tasks.terminate() is called