server/internal/client/ollama: hold DiskCache on Registry

Previously, clients of a Registry had to carry around a DiskCache to use
it. This change makes the DiskCache an optional field on the Registry
struct.

This also changes DefaultCache to initialize one on first use. This
prevents overhead of building the cache if it is never used, or per
Registry request that involves use of DefaultCache.

Also, slip in some minor docs on Trace.

Dockerfile

@@ -12,7 +12,7 @@ FROM --platform=linux/amd64 rocm/dev-almalinux-8:${ROCMVERSION}-complete AS base
 RUN yum install -y yum-utils \
     && yum-config-manager --add-repo https://dl.rockylinux.org/vault/rocky/8.5/AppStream/\$basearch/os/ \
     && rpm --import https://dl.rockylinux.org/pub/rocky/RPM-GPG-KEY-Rocky-8 \
-    && dnf install -y yum-utils ccache gcc-toolset-10-gcc-10.2.1-8.2.el8 gcc-toolset-10-gcc-c++-10.2.1-8.2.el8 gcc-toolset-10-binutils-2.35-11.el8 \
+    && dnf install -y yum-utils ccache gcc-toolset-10-gcc-10.2.1-8.2.el8 gcc-toolset-10-gcc-c++-10.2.1-8.2.el8 \
     && yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel8/x86_64/cuda-rhel8.repo
 ENV PATH=/opt/rh/gcc-toolset-10/root/usr/bin:$PATH
 
@@ -86,11 +86,10 @@ RUN --mount=type=cache,target=/root/.ccache \
         && cmake --install build --component CUDA --strip --parallel 8
 
 FROM base AS build
-WORKDIR /go/src/github.com/ollama/ollama
-COPY go.mod go.sum .
-RUN curl -fsSL https://golang.org/dl/go$(awk '/^go/ { print $2 }' go.mod).linux-$(case $(uname -m) in x86_64) echo amd64 ;; aarch64) echo arm64 ;; esac).tar.gz | tar xz -C /usr/local
+ARG GOVERSION=1.23.4
+RUN curl -fsSL https://golang.org/dl/go${GOVERSION}.linux-$(case $(uname -m) in x86_64) echo amd64 ;; aarch64) echo arm64 ;; esac).tar.gz | tar xz -C /usr/local
 ENV PATH=/usr/local/go/bin:$PATH
-RUN go mod download
+WORKDIR /go/src/github.com/ollama/ollama
 COPY . .
 ARG GOFLAGS="'-ldflags=-w -s'"
 ENV CGO_ENABLED=1

README.md

@@ -1,5 +1,5 @@
 <div align="center">
-  <a href="https://ollama.com">
+  <a href="https://ollama.com" />
     <img alt="ollama" height="200px" src="https://github.com/ollama/ollama/assets/3325447/0d0b44e2-8f4a-4e99-9b52-a5c1c741c8f7">
   </a>
 </div>
@@ -64,7 +64,7 @@ Here are some example models that can be downloaded:
 | Llama 3.1          | 8B         | 4.7GB | `ollama run llama3.1`            |
 | Llama 3.1          | 405B       | 231GB | `ollama run llama3.1:405b`       |
 | Phi 4              | 14B        | 9.1GB | `ollama run phi4`                |
-| Phi 4 Mini         | 3.8B       | 2.5GB | `ollama run phi4-mini`           |
+| Phi 3 Mini         | 3.8B       | 2.3GB | `ollama run phi3`                |
 | Gemma 2            | 2B         | 1.6GB | `ollama run gemma2:2b`           |
 | Gemma 2            | 9B         | 5.5GB | `ollama run gemma2`              |
 | Gemma 2            | 27B        | 16GB  | `ollama run gemma2:27b`          |
@@ -388,7 +388,6 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [LangBot](https://github.com/RockChinQ/LangBot) (LLM-based instant messaging bots platform, with Agents, RAG features, supports multiple platforms)
 - [1Panel](https://github.com/1Panel-dev/1Panel/) (Web-based Linux Server Management Tool)
 - [AstrBot](https://github.com/Soulter/AstrBot/) (User-friendly LLM-based multi-platform chatbot with a WebUI, supporting RAG, LLM agents, and plugins integration)
-- [Reins](https://github.com/ibrahimcetin/reins) (Easily tweak parameters, customize system prompts per chat, and enhance your AI experiments with reasoning model support.)
 
 ### Cloud
 
@@ -513,8 +512,6 @@ See the [API documentation](./docs/api.md) for all endpoints.
 - [Maid](https://github.com/Mobile-Artificial-Intelligence/maid)
 - [Ollama App](https://github.com/JHubi1/ollama-app) (Modern and easy-to-use multi-platform client for Ollama)
 - [ConfiChat](https://github.com/1runeberg/confichat) (Lightweight, standalone, multi-platform, and privacy focused LLM chat interface with optional encryption)
-- [Ollama Android Chat](https://github.com/sunshine0523/OllamaServer) (No need for Termux, start the Ollama service with one click on an Android device)
-- [Reins](https://github.com/ibrahimcetin/reins) (Easily tweak parameters, customize system prompts per chat, and enhance your AI experiments with reasoning model support.)
 
 ### Extensions & Plugins
 

cmd/cmd.go

@@ -256,7 +256,6 @@ func StopHandler(cmd *cobra.Command, args []string) error {
 		if strings.Contains(err.Error(), "not found") {
 			return fmt.Errorf("couldn't find model \"%s\" to stop", args[0])
 		}
-		return err
 	}
 	return nil
 }

docs/development.md

@@ -118,35 +118,6 @@ To run tests, use `go test`:
 go test ./...
 ```
 
-> NOTE: In rare cirumstances, you may nedd to change a package using the new
-> "synctest" package in go1.24.
->
-> If you do not have the "synctest" package enabled, you will not see build or
-> test failures resulting from your change(s), if any, locally, but CI will
-> break.
->
-> If you see failures in CI, you can either keep pushing changes to see if the
-> CI build passes, or you can enable the "synctest" package locally to see the
-> failures before pushing.
->
-> To enable the "synctest" package for testing, run the following command:
->
-> ```shell
-> GOEXPERIMENT=synctest go test ./...
-> ```
->
-> If you wish to enable synctest for all go commands, you can set the
-> `GOEXPERIMENT` environment variable in your shell profile or by using:
->
-> ```shell
-> go env -w GOEXPERIMENT=synctest
-> ```
->
-> Which will enable the "synctest" package for all go commands without needing
-> to set it for all shell sessions.
->
-> The synctest package is not required for production builds.
-
 ## Library detection
 
 Ollama looks for acceleration libraries in the following paths relative to the `ollama` executable:

llama/patches/0015-try-catch-backend-load.patch

@@ -0,0 +1,69 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Michael Yang <mxyng@pm.me>
+Date: Tue, 11 Feb 2025 14:06:36 -0800
+Subject: [PATCH] try/catch backend load
+
+---
+ ggml/src/ggml-backend-reg.cpp | 45 ++++++++++++++++++-----------------
+ 1 file changed, 23 insertions(+), 22 deletions(-)
+
+diff --git a/ggml/src/ggml-backend-reg.cpp b/ggml/src/ggml-backend-reg.cpp
+index 98d5e14d..1c19129a 100644
+--- a/ggml/src/ggml-backend-reg.cpp
++++ b/ggml/src/ggml-backend-reg.cpp
+@@ -512,32 +512,33 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
+         }
+         fs::directory_iterator dir_it(search_path, fs::directory_options::skip_permission_denied);
+         for (const auto & entry : dir_it) {
+-            if (entry.is_regular_file()) {
+-                std::wstring filename = entry.path().filename().wstring();
+-                std::wstring ext = entry.path().extension().wstring();
+-                if (filename.find(file_prefix) == 0 && ext == backend_filename_suffix()) {
+-                    dl_handle_ptr handle { dl_load_library(entry.path().wstring()) };
+-                    if (!handle && !silent) {
+-                        GGML_LOG_ERROR("%s: failed to load %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
+-                    }
+-                    if (handle) {
++            try {
++                if (entry.is_regular_file()) {
++                    std::wstring filename = entry.path().filename().wstring();
++                    std::wstring ext = entry.path().extension().wstring();
++                    if (filename.find(file_prefix) == 0 && ext == backend_filename_suffix()) {
++                        dl_handle_ptr handle { dl_load_library(entry.path().wstring()) };
++                        if (!handle) {
++                            GGML_LOG_ERROR("%s: failed to load %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
++                            continue;
++                        }
++
+                         auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
+-                        if (score_fn) {
+-                            int s = score_fn();
+-#ifndef NDEBUG
+-                            GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str(), s);
+-#endif
+-                            if (s > best_score) {
+-                                best_score = s;
+-                                best_path = entry.path().wstring();
+-                            }
+-                        } else {
+-                            if (!silent) {
+-                                GGML_LOG_INFO("%s: failed to find ggml_backend_score in %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
+-                            }
++                        if (!score_fn) {
++                            GGML_LOG_DEBUG("%s: failed to find ggml_backend_score in %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
++                            continue;
++                        }
++
++                        int s = score_fn();
++                        GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str(), s);
++                        if (s > best_score) {
++                            best_score = s;
++                            best_path = entry.path().wstring();
+                         }
+                     }
+                 }
++            } catch (const std::exception & e) {
++                GGML_LOG_ERROR("%s: failed to load %s: %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str(), e.what());
+             }
+         }
+     }

llama/patches/0016-use-std-filesystem-path-instead-of-wstring.patch

@@ -4,11 +4,11 @@ Date: Sun, 16 Feb 2025 20:00:22 -0500
 Subject: [PATCH] use std::filesystem::path instead of wstring
 
 ---
- ggml/src/ggml-backend-reg.cpp | 199 +++++++++++++++-------------------
- 1 file changed, 88 insertions(+), 111 deletions(-)
+ ggml/src/ggml-backend-reg.cpp | 144 ++++++++++++++--------------------
+ 1 file changed, 58 insertions(+), 86 deletions(-)
 
 diff --git a/ggml/src/ggml-backend-reg.cpp b/ggml/src/ggml-backend-reg.cpp
-index 98d5e14d..799af5f3 100644
+index 1c19129a..c854e6bb 100644
 --- a/ggml/src/ggml-backend-reg.cpp
 +++ b/ggml/src/ggml-backend-reg.cpp
 @@ -66,26 +66,6 @@
@@ -264,55 +264,47 @@ index 98d5e14d..799af5f3 100644
      for (const auto & search_path : search_paths) {
          if (!fs::exists(search_path)) {
              continue;
-@@ -513,29 +485,26 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
-         fs::directory_iterator dir_it(search_path, fs::directory_options::skip_permission_denied);
+@@ -514,31 +486,31 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
          for (const auto & entry : dir_it) {
-             if (entry.is_regular_file()) {
--                std::wstring filename = entry.path().filename().wstring();
--                std::wstring ext = entry.path().extension().wstring();
-+                std::string filename = entry.path().filename().string();
-+                std::string ext = entry.path().extension().string();
-                 if (filename.find(file_prefix) == 0 && ext == backend_filename_suffix()) {
--                    dl_handle_ptr handle { dl_load_library(entry.path().wstring()) };
--                    if (!handle && !silent) {
--                        GGML_LOG_ERROR("%s: failed to load %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
-+                    dl_handle_ptr handle { dl_load_library(entry.path()) };
-+                    if (!handle) {
-+                        GGML_LOG_ERROR("%s: failed to load %s\n", __func__, path_to_string(entry.path()).c_str());
-+                        continue;
-                     }
--                    if (handle) {
--                        auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
--                        if (score_fn) {
--                            int s = score_fn();
--#ifndef NDEBUG
--                            GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str(), s);
--#endif
--                            if (s > best_score) {
--                                best_score = s;
--                                best_path = entry.path().wstring();
--                            }
--                        } else {
--                            if (!silent) {
--                                GGML_LOG_INFO("%s: failed to find ggml_backend_score in %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
--                            }
--                        }
-+
-+                    auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
-+                    if (!score_fn) {
-+                        GGML_LOG_DEBUG("%s: failed to find ggml_backend_score in %s\n", __func__, path_to_string(entry.path()).c_str());
-+                        continue;
-+                    }
-+
-+                    int s = score_fn();
-+                    GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, path_to_string(entry.path()).c_str(), s);
-+                    if (s > best_score) {
-+                        best_score = s;
-+                        best_path = entry.path();
+             try {
+                 if (entry.is_regular_file()) {
+-                    std::wstring filename = entry.path().filename().wstring();
+-                    std::wstring ext = entry.path().extension().wstring();
++                    std::string filename = entry.path().filename().string();
++                    std::string ext = entry.path().extension().string();
+                     if (filename.find(file_prefix) == 0 && ext == backend_filename_suffix()) {
+-                        dl_handle_ptr handle { dl_load_library(entry.path().wstring()) };
++                        dl_handle_ptr handle { dl_load_library(entry.path()) };
+                         if (!handle) {
+-                            GGML_LOG_ERROR("%s: failed to load %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
++                            GGML_LOG_ERROR("%s: failed to load %s\n", __func__, path_to_string(entry.path()).c_str());
+                             continue;
+                         }
+ 
+                         auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
+                         if (!score_fn) {
+-                            GGML_LOG_DEBUG("%s: failed to find ggml_backend_score in %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str());
++                            GGML_LOG_DEBUG("%s: failed to find ggml_backend_score in %s\n", __func__, path_to_string(entry.path()).c_str());
+                             continue;
+                         }
+ 
+                         int s = score_fn();
+-                        GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str(), s);
++                        GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, path_to_string(entry.path()).c_str(), s);
+                         if (s > best_score) {
+                             best_score = s;
+-                            best_path = entry.path().wstring();
++                            best_path = entry.path();
+                         }
                      }
                  }
+             } catch (const std::exception & e) {
+-                GGML_LOG_ERROR("%s: failed to load %s: %s\n", __func__, utf16_to_utf8(entry.path().wstring()).c_str(), e.what());
++                GGML_LOG_ERROR("%s: failed to load %s: %s\n", __func__, path_to_string(entry.path()).c_str(), e.what());
              }
-@@ -545,7 +514,7 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
+         }
+     }
+@@ -546,7 +518,7 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
      if (best_score == 0) {
          // try to load the base backend
          for (const auto & search_path : search_paths) {
@@ -321,49 +313,3 @@ index 98d5e14d..799af5f3 100644
              if (fs::exists(path)) {
                  return get_reg().load_backend(path, silent);
              }
-@@ -560,6 +529,14 @@ void ggml_backend_load_all() {
-     ggml_backend_load_all_from_path(nullptr);
- }
- 
-+static void ggml_backend_try_load_best(const char * name, bool silent, const char * user_search_path) {
-+    try {
-+        ggml_backend_load_best(name, silent, user_search_path);
-+    } catch (const std::exception & e) {
-+        GGML_LOG_DEBUG("%s: failed to load %s: %s\n", __func__, name, e.what());
-+    }
-+}
-+
- void ggml_backend_load_all_from_path(const char * dir_path) {
- #ifdef NDEBUG
-     bool silent = true;
-@@ -567,18 +544,18 @@ void ggml_backend_load_all_from_path(const char * dir_path) {
-     bool silent = false;
- #endif
- 
--    ggml_backend_load_best("blas", silent, dir_path);
--    ggml_backend_load_best("cann", silent, dir_path);
--    ggml_backend_load_best("cuda", silent, dir_path);
--    ggml_backend_load_best("hip", silent, dir_path);
--    ggml_backend_load_best("kompute", silent, dir_path);
--    ggml_backend_load_best("metal", silent, dir_path);
--    ggml_backend_load_best("rpc", silent, dir_path);
--    ggml_backend_load_best("sycl", silent, dir_path);
--    ggml_backend_load_best("vulkan", silent, dir_path);
--    ggml_backend_load_best("opencl", silent, dir_path);
--    ggml_backend_load_best("musa", silent, dir_path);
--    ggml_backend_load_best("cpu", silent, dir_path);
-+    ggml_backend_try_load_best("blas", silent, dir_path);
-+    ggml_backend_try_load_best("cann", silent, dir_path);
-+    ggml_backend_try_load_best("cuda", silent, dir_path);
-+    ggml_backend_try_load_best("hip", silent, dir_path);
-+    ggml_backend_try_load_best("kompute", silent, dir_path);
-+    ggml_backend_try_load_best("metal", silent, dir_path);
-+    ggml_backend_try_load_best("rpc", silent, dir_path);
-+    ggml_backend_try_load_best("sycl", silent, dir_path);
-+    ggml_backend_try_load_best("vulkan", silent, dir_path);
-+    ggml_backend_try_load_best("opencl", silent, dir_path);
-+    ggml_backend_try_load_best("musa", silent, dir_path);
-+    ggml_backend_try_load_best("cpu", silent, dir_path);
-     // check the environment variable GGML_BACKEND_PATH to load an out-of-tree backend
-     const char * backend_path = std::getenv("GGML_BACKEND_PATH");
-     if (backend_path) {

ml/backend/ggml/ggml/src/ggml-backend-reg.cpp

@@ -484,29 +484,33 @@ static ggml_backend_reg_t ggml_backend_load_best(const char * name, bool silent,
         }
         fs::directory_iterator dir_it(search_path, fs::directory_options::skip_permission_denied);
         for (const auto & entry : dir_it) {
-            if (entry.is_regular_file()) {
-                std::string filename = entry.path().filename().string();
-                std::string ext = entry.path().extension().string();
-                if (filename.find(file_prefix) == 0 && ext == backend_filename_suffix()) {
-                    dl_handle_ptr handle { dl_load_library(entry.path()) };
-                    if (!handle) {
-                        GGML_LOG_ERROR("%s: failed to load %s\n", __func__, path_to_string(entry.path()).c_str());
-                        continue;
-                    }
+            try {
+                if (entry.is_regular_file()) {
+                    std::string filename = entry.path().filename().string();
+                    std::string ext = entry.path().extension().string();
+                    if (filename.find(file_prefix) == 0 && ext == backend_filename_suffix()) {
+                        dl_handle_ptr handle { dl_load_library(entry.path()) };
+                        if (!handle) {
+                            GGML_LOG_ERROR("%s: failed to load %s\n", __func__, path_to_string(entry.path()).c_str());
+                            continue;
+                        }
 
-                    auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
-                    if (!score_fn) {
-                        GGML_LOG_DEBUG("%s: failed to find ggml_backend_score in %s\n", __func__, path_to_string(entry.path()).c_str());
-                        continue;
-                    }
+                        auto score_fn = (ggml_backend_score_t) dl_get_sym(handle.get(), "ggml_backend_score");
+                        if (!score_fn) {
+                            GGML_LOG_DEBUG("%s: failed to find ggml_backend_score in %s\n", __func__, path_to_string(entry.path()).c_str());
+                            continue;
+                        }
 
-                    int s = score_fn();
-                    GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, path_to_string(entry.path()).c_str(), s);
-                    if (s > best_score) {
-                        best_score = s;
-                        best_path = entry.path();
+                        int s = score_fn();
+                        GGML_LOG_DEBUG("%s: %s score: %d\n", __func__, path_to_string(entry.path()).c_str(), s);
+                        if (s > best_score) {
+                            best_score = s;
+                            best_path = entry.path();
+                        }
                     }
                 }
+            } catch (const std::exception & e) {
+                GGML_LOG_ERROR("%s: failed to load %s: %s\n", __func__, path_to_string(entry.path()).c_str(), e.what());
             }
         }
     }
@@ -529,14 +533,6 @@ void ggml_backend_load_all() {
     ggml_backend_load_all_from_path(nullptr);
 }
 
-static void ggml_backend_try_load_best(const char * name, bool silent, const char * user_search_path) {
-    try {
-        ggml_backend_load_best(name, silent, user_search_path);
-    } catch (const std::exception & e) {
-        GGML_LOG_DEBUG("%s: failed to load %s: %s\n", __func__, name, e.what());
-    }
-}
-
 void ggml_backend_load_all_from_path(const char * dir_path) {
 #ifdef NDEBUG
     bool silent = true;
@@ -544,18 +540,18 @@ void ggml_backend_load_all_from_path(const char * dir_path) {
     bool silent = false;
 #endif
 
-    ggml_backend_try_load_best("blas", silent, dir_path);
-    ggml_backend_try_load_best("cann", silent, dir_path);
-    ggml_backend_try_load_best("cuda", silent, dir_path);
-    ggml_backend_try_load_best("hip", silent, dir_path);
-    ggml_backend_try_load_best("kompute", silent, dir_path);
-    ggml_backend_try_load_best("metal", silent, dir_path);
-    ggml_backend_try_load_best("rpc", silent, dir_path);
-    ggml_backend_try_load_best("sycl", silent, dir_path);
-    ggml_backend_try_load_best("vulkan", silent, dir_path);
-    ggml_backend_try_load_best("opencl", silent, dir_path);
-    ggml_backend_try_load_best("musa", silent, dir_path);
-    ggml_backend_try_load_best("cpu", silent, dir_path);
+    ggml_backend_load_best("blas", silent, dir_path);
+    ggml_backend_load_best("cann", silent, dir_path);
+    ggml_backend_load_best("cuda", silent, dir_path);
+    ggml_backend_load_best("hip", silent, dir_path);
+    ggml_backend_load_best("kompute", silent, dir_path);
+    ggml_backend_load_best("metal", silent, dir_path);
+    ggml_backend_load_best("rpc", silent, dir_path);
+    ggml_backend_load_best("sycl", silent, dir_path);
+    ggml_backend_load_best("vulkan", silent, dir_path);
+    ggml_backend_load_best("opencl", silent, dir_path);
+    ggml_backend_load_best("musa", silent, dir_path);
+    ggml_backend_load_best("cpu", silent, dir_path);
     // check the environment variable GGML_BACKEND_PATH to load an out-of-tree backend
     const char * backend_path = std::getenv("GGML_BACKEND_PATH");
     if (backend_path) {

scripts/install.sh

@@ -77,12 +77,11 @@ if [ -d "$OLLAMA_INSTALL_DIR/lib/ollama" ] ; then
 fi
 status "Installing ollama to $OLLAMA_INSTALL_DIR"
 $SUDO install -o0 -g0 -m755 -d $BINDIR
-$SUDO install -o0 -g0 -m755 -d "$OLLAMA_INSTALL_DIR/lib/ollama"
+$SUDO install -o0 -g0 -m755 -d "$OLLAMA_INSTALL_DIR"
 status "Downloading Linux ${ARCH} bundle"
 curl --fail --show-error --location --progress-bar \
     "https://ollama.com/download/ollama-linux-${ARCH}.tgz${VER_PARAM}" | \
     $SUDO tar -xzf - -C "$OLLAMA_INSTALL_DIR"
-
 if [ "$OLLAMA_INSTALL_DIR/bin/ollama" != "$BINDIR/ollama" ] ; then
     status "Making ollama accessible in the PATH in $BINDIR"
     $SUDO ln -sf "$OLLAMA_INSTALL_DIR/ollama" "$BINDIR/ollama"

server/internal/client/ollama/registry.go

@@ -255,6 +255,10 @@ func DefaultRegistry() (*Registry, error) {
 	}
 
 	var rc Registry
+	rc.Cache, err = defaultCache()
+	if err != nil {
+		return nil, err
+	}
 	rc.Key, err = ssh.ParseRawPrivateKey(keyPEM)
 	if err != nil {
 		return nil, err

server/internal/internal/backoff/backoff_test.go

@@ -1,5 +1,3 @@
-//go:build goexperiment.synctest
-
 package backoff
 
 import (

server/internal/internal/syncs/line_test.go

@@ -1,5 +1,3 @@
-//go:build goexperiment.synctest
-
 package syncs
 
 import (

server/internal/registry/server.go

@@ -31,10 +31,6 @@ type Local struct {
 	// Fallback, if set, is used to handle requests that are not handled by
 	// this handler.
 	Fallback http.Handler
-
-	// Prune, if set, is called to prune the local disk cache after a model
-	// is deleted.
-	Prune func() error // optional
 }
 
 // serverError is like ollama.Error, but with a Status field for the HTTP
@@ -208,10 +204,7 @@ func (s *Local) handleDelete(_ http.ResponseWriter, r *http.Request) error {
 	if !ok {
 		return &serverError{404, "not_found", "model not found"}
 	}
-	if s.Prune == nil {
-		return nil
-	}
-	return s.Prune()
+	return nil
 }
 
 func decodeUserJSON[T any](r io.Reader) (T, error) {

server/routes.go

@@ -1199,8 +1199,6 @@ func (s *Server) GenerateRoutes(rc *ollama.Registry) (http.Handler, error) {
 		Client:   rc,
 		Logger:   slog.Default(), // TODO(bmizerany): Take a logger, do not use slog.Default()
 		Fallback: r,
-
-		Prune: PruneLayers,
 	}
 
 	return rs, nil