git/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp synced 2026-04-30 11:12:04 +02:00
Code Issues Packages Projects Releases Wiki Activity

server: allow cancel loading model (#21814)

Browse Source
This commit is contained in:
Xuan-Son Nguyen 2026-04-22 00:26:09 +02:00 committed by GitHub
parent 5a4cd6741f
commit 04fe84b69d
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tools/server/server-models.cpp
View File

@ -712,6 +712,11 @@ void server_models::unload(const std::string & name) {
if (it->second.meta.is_running()) {
SRV_INF("stopping model instance name=%s\n", name.c_str());
stopping_models.insert(name);
if (it->second.meta.status == SERVER_MODEL_STATUS_LOADING) {
// special case: if model is in loading state, unloading means force-killing it
SRV_WRN("model name=%s is still loading, force-killing\n", name.c_str());
subprocess_terminate(it->second.subproc.get());
}
cv_stop.notify_all();
// status change will be handled by the managing thread
} else {