server : handle failures to restore host cache (#17078)

* server : handle failures to restore host cache * server : add tests for the prompt cache
2025-11-09 14:27:05 +02:00 · 2025-11-09 14:27:05 +02:00 · cb1adf8851
commit cb1adf8851
parent ef1d826997
2 changed files with 45 additions and 0 deletions
--- a/tools/server/server.cpp
+++ b/tools/server/server.cpp
@ -1690,6 +1690,9 @@ struct server_slot {
        bool res = prompt_cache.load(prompt, tokens, ctx, id);
        if (!res) {
            SLT_WRN(*this, "%s", "failed to load prompt from cache\n");
+
+            llama_memory_seq_rm(llama_get_memory(ctx), id, -1, -1);
+            prompt.tokens.clear();
        }
    }