diff --git a/tools/server/server-common.cpp b/tools/server/server-common.cpp
index f5a88747cb..47f5e1886d 100644
--- a/tools/server/server-common.cpp
+++ b/tools/server/server-common.cpp
@@ -1948,7 +1948,7 @@ static void server_proc_restart(char ** argv, const char * env_name, const std::
         return;
     }
 
-    setenv(env_name, env_value.c_str(), 1);
+    common_set_env(env_name, env_value);
 
     const std::string exe = server_proc_exe_path(argv);
     SRV_INF("restarting the process, exe = '%s'\n", exe.c_str());
@@ -1974,8 +1974,8 @@ void server_sleep_rst::init(int argc, char ** argv) {
 
     this->argv = argv;
 
-    const char * state = std::getenv(SLEEP_STATE_ENV);
-    if (state == nullptr) {
+    const std::string state = common_get_env(SLEEP_STATE_ENV);
+    if (state.empty()) {
         return;
     }
 
@@ -1985,12 +1985,8 @@ void server_sleep_rst::init(int argc, char ** argv) {
         SRV_ERR("failed to read the state left by the previous process: %s\n", e.what());
     }
 
-#if defined(_WIN32)
-    _putenv_s(SLEEP_STATE_ENV, "");
-#else
     // clear it now, so that child processes do not inherit it
-    unsetenv(SLEEP_STATE_ENV);
-#endif
+    common_set_env(SLEEP_STATE_ENV, "");
 }
 
 void server_sleep_rst::enable(common_params & params) {
@@ -2014,6 +2010,10 @@ void server_sleep_rst::enable(common_params & params) {
     enabled = true;
 }
 
+bool server_sleep_rst::is_boot_to_sleep() {
+    return !common_get_env(SLEEP_STATE_ENV).empty();
+}
+
 void server_sleep_rst::restart() const {
     if (!enabled) {
         return;
diff --git a/tools/server/server-common.h b/tools/server/server-common.h
index e9c823daca..9e6811ed7f 100644
--- a/tools/server/server-common.h
+++ b/tools/server/server-common.h
@@ -619,6 +619,10 @@ struct server_pipe {
 //
 
 struct server_sleep_rst {
+    // true if the process was restarted by a previous instance
+    // in this case, the model is only loaded upon the first request
+    static bool is_boot_to_sleep();
+
     // remember argv and read the state left by the previous process
     // must be called once at startup, before spawning any thread or child process
     void init(int argc, char ** argv);
@@ -626,10 +630,6 @@ struct server_sleep_rst {
     // enable the restart upon sleeping, warns and falls back to --sleep-mode free if not possible
     void enable(common_params & params);
 
-    // true if the process was restarted by a previous instance
-    // in this case, the model is only loaded upon the first request
-    bool is_boot_to_sleep() const { return !boot_state.is_null(); }
-
     // state left by the previous process, only valid if is_boot_to_sleep()
     const json & get_boot_state() const { return boot_state; }
 
diff --git a/tools/server/server.cpp b/tools/server/server.cpp
index e8030a5c20..6275cdbbca 100644
--- a/tools/server/server.cpp
+++ b/tools/server/server.cpp
@@ -132,8 +132,10 @@ int llama_server(common_params & params, int argc, char ** argv) {
     const bool is_router_server = params.model.path.empty()
                                && params.model.hf_repo.empty();
 
-    // skip device enumeration so the CUDA primary context stays uncreated
-    common_params_print_info(params, !is_router_server);
+    // may skip device enumeration so the CUDA primary context stays uncreated
+    if (!server_sleep_rst::is_boot_to_sleep()) {
+        common_params_print_info(params, !is_router_server);
+    }
 
     if (!is_router_server) {
         // validate batch size for embeddings