{"id":86009,"date":"2026-07-28T02:16:58","date_gmt":"2026-07-27T18:16:58","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/86009.html"},"modified":"2026-07-28T02:16:58","modified_gmt":"2026-07-27T18:16:58","slug":"%e3%80%8aspring-ai-%e5%a4%a7%e6%a8%a1%e5%9e%8b%e5%85%a8%e6%a0%88%e5%ae%9e%e6%88%98%e3%80%8b%e5%ad%a6%e4%b9%a0%e6%89%8b%e5%86%8c%e7%b3%bb%e5%88%97-%c2%b7-%e4%b8%93%e9%a2%98%e5%9b%9b%ef%bc%9a%e3%80%8a","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/86009.html","title":{"rendered":"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b"},"content":{"rendered":"<h2>\u4e13\u9898\u56db&#xff1a;\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18&#xff1a;\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b<\/h2>\n<h3>&#x1f4da; \u4f01\u4e1a\u7ea7 RAG \u667a\u80fd\u95ee\u7b54\u7cfb\u7edf\u5168\u6808\u5b9e\u65bd\u6307\u5357<\/h3>\n<h4>\u7b2c\u56db\u90e8\u5206&#xff1a;\u5927\u6a21\u578b\u672c\u5730\u90e8\u7f72\u4e0e\u4f18\u5316<\/h4>\n<hr \/>\n<h3>\u7b2c 1 \u7ae0 Ollama \u6982\u8ff0\u4e0e\u6838\u5fc3\u4f18\u52bf<\/h3>\n<h4>1.1 \u4ec0\u4e48\u662f Ollama&#xff1f;<\/h4>\n<p>Ollama \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u672c\u5730\u5927\u8bed\u8a00\u6a21\u578b\u8fd0\u884c\u5e73\u53f0&#xff0c;\u4e13\u4e3a\u7b80\u5316 LLM \u90e8\u7f72\u800c\u8bbe\u8ba1\u3002\u5b83\u8ba9\u5f00\u53d1\u8005\u80fd\u591f\u5728\u672c\u5730\u670d\u52a1\u5668\u4e0a\u8f7b\u677e\u8fd0\u884c\u3001\u7ba1\u7406\u548c\u4f18\u5316\u5927\u8bed\u8a00\u6a21\u578b\u3002<\/p>\n<p>\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2510<br \/>\n\u2502                      Ollama \u67b6\u6784                            \u2502<br \/>\n\u251c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2524<br \/>\n\u2502  \u7528\u6237\u8bf7\u6c42 \u2192 API \u63a5\u53e3 \u2192 \u6a21\u578b\u52a0\u8f7d \u2192 \u63a8\u7406\u5f15\u64ce \u2192 \u6d41\u5f0f\u54cd\u5e94        \u2502<br \/>\n\u2502                      (llama.cpp \u540e\u7aef)                        \u2502<br \/>\n\u2514\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2518<\/p>\n<h4>1.2 \u4e3a\u4ec0\u4e48\u9009\u62e9 Ollama&#xff1f;<\/h4>\n<table>\n<tr>\u5bf9\u6bd4\u9879Ollama\u539f\u751f llama.cppHuggingFacevLLM<\/tr>\n<tbody>\n<tr>\n<td>\u90e8\u7f72\u96be\u5ea6<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50\u2b50<\/td>\n<td>\u2b50\u2b50\u2b50<\/td>\n<td>\u2b50\u2b50<\/td>\n<td>\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<tr>\n<td>\u6a21\u578b\u7ba1\u7406<\/td>\n<td>\u4e00\u952e\u62c9\u53d6<\/td>\n<td>\u624b\u52a8\u4e0b\u8f7d<\/td>\n<td>\u624b\u52a8\u4e0b\u8f7d<\/td>\n<td>\u624b\u52a8\u914d\u7f6e<\/td>\n<\/tr>\n<tr>\n<td>\u5185\u5b58\u4f18\u5316<\/td>\n<td>\u81ea\u52a8\u91cf\u5316<\/td>\n<td>\u9700\u624b\u52a8\u914d\u7f6e<\/td>\n<td>\u9700\u624b\u52a8\u914d\u7f6e<\/td>\n<td>\u9700\u624b\u52a8\u914d\u7f6e<\/td>\n<\/tr>\n<tr>\n<td>API \u63a5\u53e3<\/td>\n<td>RESTful<\/td>\n<td>\u9700\u81ea\u884c\u5c01\u88c5<\/td>\n<td>\u9700\u81ea\u884c\u5c01\u88c5<\/td>\n<td>RESTful<\/td>\n<\/tr>\n<tr>\n<td>\u591a\u6a21\u578b\u652f\u6301<\/td>\n<td>100&#043;<\/td>\n<td>\u4f9d\u8d56 GGUF<\/td>\n<td>\u4f9d\u8d56 PyTorch<\/td>\n<td>\u4f9d\u8d56 PyTorch<\/td>\n<\/tr>\n<tr>\n<td>\u793e\u533a\u751f\u6001<\/td>\n<td>\u6d3b\u8dc3<\/td>\n<td>\u6d3b\u8dc3<\/td>\n<td>\u6700\u6d3b\u8dc3<\/td>\n<td>\u6d3b\u8dc3<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>1.3 Ollama \u6838\u5fc3\u7279\u6027<\/h4>\n<table>\n<tr>\u7279\u6027\u8bf4\u660e\u6280\u672f\u5b9e\u73b0\u6027\u80fd\u63d0\u5347<\/tr>\n<tbody>\n<tr>\n<td>&#x1f680; \u4e00\u952e\u90e8\u7f72<\/td>\n<td>ollama pull &lt;model&gt;<\/td>\n<td>\u81ea\u52a8\u4e0b\u8f7d\u91cf\u5316\u6a21\u578b<\/td>\n<td>\u90e8\u7f72\u65f6\u95f4\u4ece 30 \u5206\u949f\u219230 \u79d2<\/td>\n<\/tr>\n<tr>\n<td>&#x1f4e6; \u6a21\u578b\u91cf\u5316<\/td>\n<td>Q4\/Q5\/Q8 \u591a\u79cd\u7cbe\u5ea6<\/td>\n<td>GGUF \u683c\u5f0f<\/td>\n<td>\u5185\u5b58\u5360\u7528\u964d\u4f4e 50-75%<\/td>\n<\/tr>\n<tr>\n<td>&#x1f50c; REST API<\/td>\n<td>\u6807\u51c6 HTTP \u63a5\u53e3<\/td>\n<td>\/api\/generate<\/td>\n<td>\u96c6\u6210\u6548\u7387\u63d0\u5347 80%<\/td>\n<\/tr>\n<tr>\n<td>&#x1f504; \u6d41\u5f0f\u8f93\u51fa<\/td>\n<td>SSE \u534f\u8bae<\/td>\n<td>stream: true<\/td>\n<td>\u9996\u5b57\u5ef6\u8fdf&lt;500ms<\/td>\n<\/tr>\n<tr>\n<td>&#x1f4be; \u6a21\u578b\u7f13\u5b58<\/td>\n<td>\u81ea\u52a8\u7ba1\u7406<\/td>\n<td>\u672c\u5730\u5b58\u50a8<\/td>\n<td>\u4e8c\u6b21\u52a0\u8f7d\u901f\u5ea6\u63d0\u5347 10 \u500d<\/td>\n<\/tr>\n<tr>\n<td>&#x1f6e1; \u5b89\u5168\u9694\u79bb<\/td>\n<td>\u5bb9\u5668\u5316\u8fd0\u884c<\/td>\n<td>Docker \u652f\u6301<\/td>\n<td>\u751f\u4ea7\u73af\u5883\u5c31\u7eea<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<hr \/>\n<h3>\u7b2c 2 \u7ae0 Ollama \u5b89\u88c5\u4e0e\u914d\u7f6e\u5b9e\u6218<\/h3>\n<h4>2.1 Linux \u4e00\u952e\u5b89\u88c5\u811a\u672c<\/h4>\n<p><span class=\"token shebang important\">#!\/bin\/bash<\/span><br \/>\n<span class=\"token comment\"># install-ollama.sh<\/span><\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;&#x1f680; \u5f00\u59cb\u5b89\u88c5 Ollama&#8230;&#034;<\/span><\/p>\n<p><span class=\"token comment\"># 1. \u4e0b\u8f7d\u5e76\u6267\u884c\u5b98\u65b9\u5b89\u88c5\u811a\u672c<\/span><br \/>\n<span class=\"token function\">curl<\/span> <span class=\"token parameter variable\">-fsSL<\/span> https:\/\/ollama.com\/install.sh <span class=\"token operator\">|<\/span> <span class=\"token function\">sh<\/span><\/p>\n<p><span class=\"token comment\"># 2. \u9a8c\u8bc1\u5b89\u88c5<\/span><br \/>\nollama <span class=\"token parameter variable\">&#8211;version<\/span><\/p>\n<p><span class=\"token comment\"># 3. \u521b\u5efa\u7cfb\u7edf\u670d\u52a1<\/span><br \/>\n<span class=\"token function\">sudo<\/span> systemctl <span class=\"token builtin class-name\">enable<\/span> ollama<br \/>\n<span class=\"token function\">sudo<\/span> systemctl start ollama<\/p>\n<p><span class=\"token comment\"># 4. \u68c0\u67e5\u670d\u52a1\u72b6\u6001<\/span><br \/>\n<span class=\"token function\">sudo<\/span> systemctl status ollama<\/p>\n<p><span class=\"token comment\"># 5. \u914d\u7f6e\u5f00\u673a\u81ea\u542f<\/span><br \/>\n<span class=\"token function\">sudo<\/span> systemctl <span class=\"token builtin class-name\">enable<\/span> ollama<\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;\u2705 Ollama \u5b89\u88c5\u5b8c\u6210&#xff01;&#034;<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;   &#8211; \u670d\u52a1\u7aef\u53e3&#xff1a;11434&#034;<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;   &#8211; \u6a21\u578b\u76ee\u5f55&#xff1a;~\/.ollama\/models&#034;<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;   &#8211; \u65e5\u5fd7\u4f4d\u7f6e&#xff1a;journalctl -u ollama&#034;<\/span><\/p>\n<h4>2.2 Docker \u90e8\u7f72\u65b9\u6848&#xff08;\u63a8\u8350\u751f\u4ea7\u73af\u5883&#xff09;<\/h4>\n<p><span class=\"token comment\"># docker-compose.yml<\/span><br \/>\n<span class=\"token key atrule\">services<\/span><span class=\"token punctuation\">:<\/span><br \/>\n  <span class=\"token key atrule\">ollama<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token key atrule\">container_name<\/span><span class=\"token punctuation\">:<\/span> ollama<br \/>\n    <span class=\"token key atrule\">image<\/span><span class=\"token punctuation\">:<\/span> ollama\/ollama<span class=\"token punctuation\">:<\/span>latest<br \/>\n    <span class=\"token key atrule\">ports<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token string\">&#034;11434:11434&#034;<\/span><br \/>\n    <span class=\"token key atrule\">volumes<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> .\/ollama<span class=\"token punctuation\">&#8211;<\/span>data<span class=\"token punctuation\">:<\/span>\/root\/.ollama<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> .\/models<span class=\"token punctuation\">:<\/span>\/models<br \/>\n    <span class=\"token key atrule\">environment<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_HOST&#061;0.0.0.0<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_PORT&#061;11434<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_KEEP_ALIVE&#061;5m<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_NUM_THREADS&#061;2<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_MAX_LOADED_MODELS&#061;1<br \/>\n    <span class=\"token key atrule\">deploy<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token key atrule\">resources<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token key atrule\">limits<\/span><span class=\"token punctuation\">:<\/span><br \/>\n          <span class=\"token key atrule\">memory<\/span><span class=\"token punctuation\">:<\/span> 2G<br \/>\n          <span class=\"token key atrule\">cpus<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#039;2.0&#039;<\/span><br \/>\n    <span class=\"token key atrule\">restart<\/span><span class=\"token punctuation\">:<\/span> unless<span class=\"token punctuation\">&#8211;<\/span>stopped<br \/>\n    <span class=\"token key atrule\">healthcheck<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token key atrule\">test<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;CMD&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;curl&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;-f&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;http:\/\/localhost:11434\/api\/tags&#034;<\/span><span class=\"token punctuation\">]<\/span><br \/>\n      <span class=\"token key atrule\">interval<\/span><span class=\"token punctuation\">:<\/span> 30s<br \/>\n      <span class=\"token key atrule\">timeout<\/span><span class=\"token punctuation\">:<\/span> 10s<br \/>\n      <span class=\"token key atrule\">retries<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token number\">3<\/span><br \/>\n    <span class=\"token key atrule\">networks<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> rag<span class=\"token punctuation\">&#8211;<\/span>network<\/p>\n<p><span class=\"token key atrule\">networks<\/span><span class=\"token punctuation\">:<\/span><br \/>\n  <span class=\"token key atrule\">rag-network<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token key atrule\">driver<\/span><span class=\"token punctuation\">:<\/span> bridge<\/p>\n<h4>2.3 \u5173\u952e\u73af\u5883\u53d8\u91cf\u914d\u7f6e<\/h4>\n<table>\n<tr>\u53d8\u91cf\u540d\u9ed8\u8ba4\u503c\u63a8\u8350\u503c (2C4G)\u8bf4\u660e\u5f71\u54cd<\/tr>\n<tbody>\n<tr>\n<td>OLLAMA_HOST<\/td>\n<td>127.0.0.1<\/td>\n<td>0.0.0.0<\/td>\n<td>\u76d1\u542c\u5730\u5740<\/td>\n<td>\u5141\u8bb8\u5916\u90e8\u8bbf\u95ee<\/td>\n<\/tr>\n<tr>\n<td>OLLAMA_PORT<\/td>\n<td>11434<\/td>\n<td>11434<\/td>\n<td>\u670d\u52a1\u7aef\u53e3<\/td>\n<td>\u9632\u706b\u5899\u9700\u653e\u884c<\/td>\n<\/tr>\n<tr>\n<td>OLLAMA_KEEP_ALIVE<\/td>\n<td>5m<\/td>\n<td>5m<\/td>\n<td>\u6a21\u578b\u4fdd\u6301\u65f6\u95f4<\/td>\n<td>\u8d85\u65f6\u81ea\u52a8\u91ca\u653e\u5185\u5b58<\/td>\n<\/tr>\n<tr>\n<td>OLLAMA_NUM_THREADS<\/td>\n<td>CPU \u6838\u5fc3\u6570<\/td>\n<td>2<\/td>\n<td>\u63a8\u7406\u7ebf\u7a0b\u6570<\/td>\n<td>\u9632\u6b62 CPU \u6ee1\u8f7d<\/td>\n<\/tr>\n<tr>\n<td>OLLAMA_MAX_LOADED_MODELS<\/td>\n<td>1<\/td>\n<td>1<\/td>\n<td>\u6700\u5927\u52a0\u8f7d\u6a21\u578b\u6570<\/td>\n<td>\u9632\u6b62\u5185\u5b58\u6ea2\u51fa<\/td>\n<\/tr>\n<tr>\n<td>OLLAMA_MODELS<\/td>\n<td>~\/.ollama\/models<\/td>\n<td>\/models<\/td>\n<td>\u6a21\u578b\u5b58\u50a8\u8def\u5f84<\/td>\n<td>\u4fbf\u4e8e\u6570\u636e\u6301\u4e45\u5316<\/td>\n<\/tr>\n<tr>\n<td>OLLAMA_DEBUG<\/td>\n<td>false<\/td>\n<td>false<\/td>\n<td>\u8c03\u8bd5\u6a21\u5f0f<\/td>\n<td>\u751f\u4ea7\u73af\u5883\u5173\u95ed<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>2.4 30 \u79d2\u9a8c\u8bc1\u5b89\u88c5\u6210\u529f<\/h4>\n<p><span class=\"token comment\"># 1. \u68c0\u67e5\u670d\u52a1\u72b6\u6001<\/span><br \/>\n<span class=\"token function\">curl<\/span> http:\/\/localhost:11434\/api\/tags<\/p>\n<p><span class=\"token comment\"># 2. \u62c9\u53d6\u6d4b\u8bd5\u6a21\u578b<\/span><br \/>\nollama pull qwen2.5:0.5b<\/p>\n<p><span class=\"token comment\"># 3. \u6d4b\u8bd5\u63a8\u7406<\/span><br \/>\nollama run qwen2.5:0.5b <span class=\"token string\">&#034;\u4f60\u597d&#xff0c;\u8bf7\u81ea\u6211\u4ecb\u7ecd&#034;<\/span><\/p>\n<p><span class=\"token comment\"># 4. \u6d4b\u8bd5 API \u63a5\u53e3<\/span><br \/>\n<span class=\"token function\">curl<\/span> http:\/\/localhost:11434\/api\/generate <span class=\"token parameter variable\">-d<\/span> <span class=\"token string\">&#039;{<br \/>\n  &#034;model&#034;: &#034;qwen2.5:0.5b&#034;,<br \/>\n  &#034;prompt&#034;: &#034;\u4ec0\u4e48\u662f RAG \u6280\u672f&#xff1f;&#034;,<br \/>\n  &#034;stream&#034;: false<br \/>\n}&#039;<\/span> <span class=\"token operator\">|<\/span> jq <span class=\"token string\">&#039;.response&#039;<\/span><\/p>\n<p><span class=\"token comment\"># 5. \u67e5\u770b\u5df2\u5b89\u88c5\u6a21\u578b<\/span><br \/>\nollama list<\/p>\n<p><span class=\"token comment\"># 6. \u67e5\u770b\u6a21\u578b\u8be6\u7ec6\u4fe1\u606f<\/span><br \/>\nollama show qwen2.5:0.5b<\/p>\n<hr \/>\n<h3>\u7b2c 3 \u7ae0 \u6a21\u578b\u91cf\u5316\u4e0e\u9009\u578b\u6307\u5357<\/h3>\n<h4>3.1 \u91cf\u5316\u6280\u672f\u8be6\u89e3<\/h4>\n<p>\u91cf\u5316\u662f\u5c06\u6a21\u578b\u6743\u91cd\u4ece\u9ad8\u7cbe\u5ea6&#xff08;FP32&#xff09;\u8f6c\u6362\u4e3a\u4f4e\u7cbe\u5ea6&#xff08;INT4\/INT8&#xff09;\u7684\u6280\u672f&#xff0c;\u53ef\u5927\u5e45\u964d\u4f4e\u5185\u5b58\u5360\u7528\u3002<\/p>\n<h5>\u91cf\u5316\u7b49\u7ea7\u5bf9\u6bd4<\/h5>\n<table>\n<tr>\u91cf\u5316\u7c7b\u578b\u7cbe\u5ea6\u5185\u5b58\u538b\u7f29\u6bd4\u7cbe\u5ea6\u635f\u5931\u9002\u7528\u573a\u666f\u6587\u4ef6\u5927\u5c0f\u793a\u4f8b (7B \u6a21\u578b)<\/tr>\n<tbody>\n<tr>\n<td>FP32<\/td>\n<td>32 \u4f4d\u6d6e\u70b9<\/td>\n<td>1x<\/td>\n<td>0%<\/td>\n<td>\u8bad\u7ec3\/\u7814\u7a76<\/td>\n<td>28GB<\/td>\n<\/tr>\n<tr>\n<td>FP16<\/td>\n<td>16 \u4f4d\u6d6e\u70b9<\/td>\n<td>2x<\/td>\n<td>&lt;1%<\/td>\n<td>\u9ad8\u7cbe\u5ea6\u63a8\u7406<\/td>\n<td>14GB<\/td>\n<\/tr>\n<tr>\n<td>Q8_0<\/td>\n<td>8 \u4f4d\u6574\u6570<\/td>\n<td>4x<\/td>\n<td>1-2%<\/td>\n<td>\u751f\u4ea7\u73af\u5883\u63a8\u8350<\/td>\n<td>7GB<\/td>\n<\/tr>\n<tr>\n<td>Q5_K_M<\/td>\n<td>5 \u4f4d\u6df7\u5408<\/td>\n<td>5.5x<\/td>\n<td>2-3%<\/td>\n<td>\u5e73\u8861\u65b9\u6848<\/td>\n<td>5GB<\/td>\n<\/tr>\n<tr>\n<td>Q4_K_M<\/td>\n<td>4 \u4f4d\u6df7\u5408<\/td>\n<td>7x<\/td>\n<td>3-5%<\/td>\n<td>\u4f4e\u914d\u670d\u52a1\u5668\u63a8\u8350<\/td>\n<td>4GB<\/td>\n<\/tr>\n<tr>\n<td>Q3_K_S<\/td>\n<td>3 \u4f4d\u6df7\u5408<\/td>\n<td>9x<\/td>\n<td>5-8%<\/td>\n<td>\u6781\u7aef\u4f4e\u914d<\/td>\n<td>3GB<\/td>\n<\/tr>\n<tr>\n<td>Q2_K<\/td>\n<td>2 \u4f4d\u6df7\u5408<\/td>\n<td>12x<\/td>\n<td>8-12%<\/td>\n<td>\u4e0d\u63a8\u8350<\/td>\n<td>2.5GB<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>3.2 Ollama \u6a21\u578b\u547d\u540d\u89c4\u5219<\/h4>\n<p>&lt;\u6a21\u578b\u540d&gt;:&lt;\u53c2\u6570\u91cf&gt;&lt;\u91cf\u5316\u7248\u672c&gt;<\/p>\n<p>\u793a\u4f8b&#xff1a;<br \/>\n&#8211; qwen2.5:0.5b        \u2192 Qwen2.5 0.5B \u53c2\u6570&#xff0c;\u9ed8\u8ba4\u91cf\u5316<br \/>\n&#8211; qwen2.5:1.5b-q4_K_M \u2192 Qwen2.5 1.5B \u53c2\u6570&#xff0c;Q4 \u91cf\u5316<br \/>\n&#8211; qwen2.5:7b-q8_0     \u2192 Qwen2.5 7B \u53c2\u6570&#xff0c;Q8 \u91cf\u5316<br \/>\n&#8211; bge-m3:latest       \u2192 BGE-M3 Embedding \u6a21\u578b&#xff0c;\u6700\u65b0\u7248<\/p>\n<h4>3.3 2025 \u5e74\u63a8\u8350\u6a21\u578b\u6e05\u5355<\/h4>\n<table>\n<tr>\u573a\u666f\u6a21\u578b\u53c2\u6570\u91cf\u91cf\u5316\u7248\u672c\u5185\u5b58\u5360\u7528\u63a8\u7406\u901f\u5ea6\u63a8\u8350\u6307\u6570<\/tr>\n<tbody>\n<tr>\n<td>2C4G \u670d\u52a1\u5668<\/td>\n<td>qwen2.5:0.5b<\/td>\n<td>0.5B<\/td>\n<td>Q4_K_M<\/td>\n<td>0.8GB<\/td>\n<td>50 token\/s<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<tr>\n<td>4C8G \u670d\u52a1\u5668<\/td>\n<td>qwen2.5:1.5b<\/td>\n<td>1.5B<\/td>\n<td>Q4_K_M<\/td>\n<td>1.6GB<\/td>\n<td>35 token\/s<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<tr>\n<td>8C16G \u670d\u52a1\u5668<\/td>\n<td>qwen2.5:7b<\/td>\n<td>7B<\/td>\n<td>Q4_K_M<\/td>\n<td>4.5GB<\/td>\n<td>20 token\/s<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<tr>\n<td>GPU \u73af\u5883<\/td>\n<td>qwen2.5:7b<\/td>\n<td>7B<\/td>\n<td>FP16<\/td>\n<td>14GB<\/td>\n<td>80 token\/s<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<tr>\n<td>Embedding<\/td>\n<td>bge-m3<\/td>\n<td>567M<\/td>\n<td>Q4_K_M<\/td>\n<td>1.2GB<\/td>\n<td>100 \u53e5\/s<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<tr>\n<td>\u4ee3\u7801\u751f\u6210<\/td>\n<td>codellama:7b<\/td>\n<td>7B<\/td>\n<td>Q4_K_M<\/td>\n<td>4.5GB<\/td>\n<td>18 token\/s<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<tr>\n<td>\u591a\u8bed\u8a00<\/td>\n<td>llama3.1:8b<\/td>\n<td>8B<\/td>\n<td>Q4_K_M<\/td>\n<td>5.0GB<\/td>\n<td>18 token\/s<\/td>\n<td>\u2b50\u2b50\u2b50\u2b50<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>3.4 \u6a21\u578b\u62c9\u53d6\u4e0e\u7ba1\u7406\u547d\u4ee4<\/h4>\n<p><span class=\"token comment\"># 1. \u62c9\u53d6\u6a21\u578b<\/span><br \/>\nollama pull qwen2.5:1.5b<\/p>\n<p><span class=\"token comment\"># 2. \u62c9\u53d6\u7279\u5b9a\u91cf\u5316\u7248\u672c<\/span><br \/>\nollama pull qwen2.5:1.5b-q4_K_M<\/p>\n<p><span class=\"token comment\"># 3. \u67e5\u770b\u5df2\u5b89\u88c5\u6a21\u578b<\/span><br \/>\nollama list<\/p>\n<p><span class=\"token comment\"># 4. \u67e5\u770b\u6a21\u578b\u8be6\u7ec6\u4fe1\u606f<\/span><br \/>\nollama show qwen2.5:1.5b<\/p>\n<p><span class=\"token comment\"># 5. \u67e5\u770b\u6a21\u578b\u6587\u4ef6\u4f4d\u7f6e<\/span><br \/>\n<span class=\"token function\">ls<\/span> <span class=\"token parameter variable\">-lh<\/span> ~\/.ollama\/models<\/p>\n<p><span class=\"token comment\"># 6. \u5220\u9664\u6a21\u578b<\/span><br \/>\nollama <span class=\"token function\">rm<\/span> qwen2.5:1.5b<\/p>\n<p><span class=\"token comment\"># 7. \u590d\u5236\u6a21\u578b&#xff08;\u521b\u5efa\u81ea\u5b9a\u4e49\u7248\u672c&#xff09;<\/span><br \/>\nollama <span class=\"token function\">cp<\/span> qwen2.5:1.5b my-qwen<\/p>\n<p><span class=\"token comment\"># 8. \u5bfc\u51fa\u6a21\u578b<\/span><br \/>\nollama <span class=\"token function\">cp<\/span> qwen2.5:1.5b \/backup\/my-model<\/p>\n<p><span class=\"token comment\"># 9. \u6e05\u7406\u672a\u4f7f\u7528\u6a21\u578b<\/span><br \/>\nollama prune<\/p>\n<hr \/>\n<h3>\u7b2c 4 \u7ae0 2G \u5185\u5b58\u6781\u9650\u4f18\u5316\u5b9e\u6218<\/h3>\n<h4>4.1 \u5185\u5b58\u5206\u914d\u7b56\u7565&#xff08;2C4G \u670d\u52a1\u5668&#xff09;<\/h4>\n<p>\u26a0\ufe0f \u8b66\u544a&#xff1a;2G \u5185\u5b58\u662f RAG \u7cfb\u7edf\u7684&#034;\u751f\u5b58\u7ebf&#034;&#xff0c;\u5fc5\u987b\u6267\u884c\u7cbe\u7ec6\u5316\u8d44\u6e90\u7ba1\u7406<\/p>\n<h5>\u5185\u5b58\u8d44\u4ea7\u8d1f\u503a\u8868<\/h5>\n<table>\n<tr>\u7ec4\u4ef6\u7269\u7406\u5185\u5b58\u4f18\u5316\u540e\u4f18\u5316\u624b\u6bb5\u98ce\u9669\u7b49\u7ea7<\/tr>\n<tbody>\n<tr>\n<td>Ollama (LLM)<\/td>\n<td>2.5GB<\/td>\n<td>1.0GB<\/td>\n<td>Q4 \u91cf\u5316 &#043; \u7ebf\u7a0b\u9650\u5236<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>Ollama (Embed)<\/td>\n<td>1.2GB<\/td>\n<td>0.4GB<\/td>\n<td>\u6362\u7528 bge-small-zh<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>Milvus<\/td>\n<td>2.8GB<\/td>\n<td>1.0GB<\/td>\n<td>\u5185\u5b58\u9650\u5236\u914d\u7f6e<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>Spring Boot<\/td>\n<td>0.8GB<\/td>\n<td>0.4GB<\/td>\n<td>JVM \u53c2\u6570\u9650\u5236<\/td>\n<td>&#x1f7e1; \u4e2d<\/td>\n<\/tr>\n<tr>\n<td>\u7cfb\u7edf\u9884\u7559<\/td>\n<td>0.7GB<\/td>\n<td>0.2GB<\/td>\n<td>\u5173\u95ed\u5197\u4f59\u670d\u52a1<\/td>\n<td>&#x1f7e1; \u4e2d<\/td>\n<\/tr>\n<tr>\n<td>\u603b\u8ba1<\/td>\n<td>8.0GB<\/td>\n<td>3.0GB<\/td>\n<td>\u9700 Swap \u8865\u5145<\/td>\n<td>&#8211;<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>4.2 Swap \u865a\u62df\u5185\u5b58\u914d\u7f6e&#xff08;\u5fc5\u505a&#xff09;<\/h4>\n<p><span class=\"token shebang important\">#!\/bin\/bash<\/span><br \/>\n<span class=\"token comment\"># setup-swap.sh<\/span><\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;&#x1f680; \u914d\u7f6e Swap \u865a\u62df\u5185\u5b58&#8230;&#034;<\/span><\/p>\n<p><span class=\"token comment\"># 1. \u68c0\u67e5\u73b0\u6709 Swap<\/span><br \/>\n<span class=\"token function\">free<\/span> <span class=\"token parameter variable\">-h<\/span><\/p>\n<p><span class=\"token comment\"># 2. \u521b\u5efa 4G Swap \u6587\u4ef6&#xff08;2G \u5185\u5b58\u670d\u52a1\u5668\u5efa\u8bae 4-8G Swap&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> fallocate <span class=\"token parameter variable\">-l<\/span> 4G \/swapfile<\/p>\n<p><span class=\"token comment\"># 3. \u8bbe\u7f6e\u6743\u9650&#xff08;\u5b89\u5168\u8981\u6c42&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">chmod<\/span> <span class=\"token number\">600<\/span> \/swapfile<\/p>\n<p><span class=\"token comment\"># 4. \u683c\u5f0f\u5316\u4e3a Swap<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">mkswap<\/span> \/swapfile<\/p>\n<p><span class=\"token comment\"># 5. \u542f\u7528 Swap<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">swapon<\/span> \/swapfile<\/p>\n<p><span class=\"token comment\"># 6. \u9a8c\u8bc1 Swap<\/span><br \/>\n<span class=\"token function\">free<\/span> <span class=\"token parameter variable\">-h<\/span><\/p>\n<p><span class=\"token comment\"># 7. \u6c38\u4e45\u751f\u6548&#xff08;\u5199\u5165 fstab&#xff09;<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#039;\/swapfile none swap sw 0 0&#039;<\/span> <span class=\"token operator\">|<\/span> <span class=\"token function\">sudo<\/span> <span class=\"token function\">tee<\/span> <span class=\"token parameter variable\">-a<\/span> \/etc\/fstab<\/p>\n<p><span class=\"token comment\"># 8. \u8c03\u6574 Swap \u4f7f\u7528\u503e\u5411&#xff08;\u9ed8\u8ba4 60&#xff0c;\u6539\u4e3a 30 \u51cf\u5c11 Swap \u4f7f\u7528&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">sysctl<\/span> <span class=\"token assign-left variable\">vm.swappiness<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">30<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#039;vm.swappiness&#061;30&#039;<\/span> <span class=\"token operator\">|<\/span> <span class=\"token function\">sudo<\/span> <span class=\"token function\">tee<\/span> <span class=\"token parameter variable\">-a<\/span> \/etc\/sysctl.conf<\/p>\n<p><span class=\"token comment\"># 9. \u8c03\u6574\u810f\u9875\u56de\u5199\u7b56\u7565<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">sysctl<\/span> <span class=\"token assign-left variable\">vm.dirty_ratio<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">10<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">sysctl<\/span> <span class=\"token assign-left variable\">vm.dirty_background_ratio<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">5<\/span><\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;\u2705 Swap \u914d\u7f6e\u5b8c\u6210&#xff01;&#034;<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;   &#8211; Swap \u5927\u5c0f&#xff1a;4GB&#034;<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;   &#8211; Swappiness: 30&#xff08;\u51cf\u5c11 Swap \u4f7f\u7528\u9891\u7387&#xff09;&#034;<\/span><\/p>\n<h4>4.3 Ollama \u542f\u52a8\u53c2\u6570\u4f18\u5316<\/h4>\n<p><span class=\"token comment\"># 2C4G \u670d\u52a1\u5668\u63a8\u8350\u542f\u52a8\u547d\u4ee4<\/span><br \/>\nollama serve <span class=\"token punctuation\">\\\\<\/span><br \/>\n  <span class=\"token parameter variable\">&#8211;host<\/span> <span class=\"token number\">0.0<\/span>.0.0 <span class=\"token punctuation\">\\\\<\/span><br \/>\n  <span class=\"token parameter variable\">&#8211;port<\/span> <span class=\"token number\">11434<\/span> <span class=\"token punctuation\">\\\\<\/span><br \/>\n  &#8211;num-gpu <span class=\"token number\">0<\/span> <span class=\"token punctuation\">\\\\<\/span><br \/>\n  &#8211;num-threads <span class=\"token number\">2<\/span> <span class=\"token punctuation\">\\\\<\/span><br \/>\n  &#8211;max-loaded-models <span class=\"token number\">1<\/span><\/p>\n<p><span class=\"token comment\"># \u73af\u5883\u53d8\u91cf\u65b9\u5f0f&#xff08;\u63a8\u8350 Docker \u4f7f\u7528&#xff09;<\/span><br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">OLLAMA_HOST<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">0.0<\/span>.0.0<br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">OLLAMA_PORT<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">11434<\/span><br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">OLLAMA_NUM_THREADS<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">2<\/span><br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">OLLAMA_MAX_LOADED_MODELS<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">1<\/span><br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">OLLAMA_KEEP_ALIVE<\/span><span class=\"token operator\">&#061;<\/span>5m<br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">OLLAMA_DEBUG<\/span><span class=\"token operator\">&#061;<\/span>false<\/p>\n<p>ollama serve<\/p>\n<h4>4.4 \u6a21\u578b\u52a0\u8f7d\u4f18\u5316\u6280\u5de7<\/h4>\n<p><span class=\"token comment\"># 1. \u9884\u52a0\u8f7d\u6a21\u578b\u5230\u5185\u5b58&#xff08;\u51cf\u5c11\u9996\u6b21\u8bf7\u6c42\u5ef6\u8fdf&#xff09;<\/span><br \/>\n<span class=\"token function\">curl<\/span> http:\/\/localhost:11434\/api\/generate <span class=\"token parameter variable\">-d<\/span> <span class=\"token string\">&#039;{<br \/>\n  &#034;model&#034;: &#034;qwen2.5:0.5b&#034;,<br \/>\n  &#034;prompt&#034;: &#034;warmup&#034;,<br \/>\n  &#034;stream&#034;: false<br \/>\n}&#039;<\/span><\/p>\n<p><span class=\"token comment\"># 2. \u8bbe\u7f6e\u6a21\u578b\u4fdd\u6301\u65f6\u95f4&#xff08;5 \u5206\u949f\u540e\u81ea\u52a8\u91ca\u653e&#xff09;<\/span><br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">OLLAMA_KEEP_ALIVE<\/span><span class=\"token operator\">&#061;<\/span>5m<\/p>\n<p><span class=\"token comment\"># 3. \u9650\u5236\u5e76\u53d1\u8bf7\u6c42\u6570&#xff08;\u9632\u6b62\u5185\u5b58\u6ea2\u51fa&#xff09;<\/span><br \/>\n<span class=\"token comment\"># \u5728 Modelfile \u4e2d\u8bbe\u7f6e<\/span><br \/>\nPARAMETER num_thread <span class=\"token number\">2<\/span><br \/>\nPARAMETER max_queue <span class=\"token number\">1<\/span><\/p>\n<p><span class=\"token comment\"># 4. \u76d1\u63a7\u6a21\u578b\u5185\u5b58\u5360\u7528<\/span><br \/>\n<span class=\"token function\">watch<\/span> <span class=\"token parameter variable\">-n<\/span> <span class=\"token number\">1<\/span> <span class=\"token string\">&#039;ollama ps&#039;<\/span><\/p>\n<p><span class=\"token comment\"># 5. \u624b\u52a8\u5378\u8f7d\u6a21\u578b<\/span><br \/>\n<span class=\"token function\">curl<\/span> http:\/\/localhost:11434\/api\/generate <span class=\"token parameter variable\">-d<\/span> <span class=\"token string\">&#039;{<br \/>\n  &#034;model&#034;: &#034;qwen2.5:0.5b&#034;,<br \/>\n  &#034;prompt&#034;: &#034;&#034;,<br \/>\n  &#034;keep_alive&#034;: 0<br \/>\n}&#039;<\/span><\/p>\n<h4>4.5 \u7cfb\u7edf\u7ea7\u4f18\u5316\u914d\u7f6e<\/h4>\n<p><span class=\"token shebang important\">#!\/bin\/bash<\/span><br \/>\n<span class=\"token comment\"># system-optimize.sh<\/span><\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;&#x1f680; \u7cfb\u7edf\u7ea7\u4f18\u5316\u914d\u7f6e&#8230;&#034;<\/span><\/p>\n<p><span class=\"token comment\"># 1. \u5173\u95ed\u4e0d\u5fc5\u8981\u7684\u7cfb\u7edf\u670d\u52a1<\/span><br \/>\n<span class=\"token function\">sudo<\/span> systemctl stop bluetooth<br \/>\n<span class=\"token function\">sudo<\/span> systemctl disable bluetooth<br \/>\n<span class=\"token function\">sudo<\/span> systemctl stop cups<br \/>\n<span class=\"token function\">sudo<\/span> systemctl disable cups<\/p>\n<p><span class=\"token comment\"># 2. \u6e05\u7406\u7cfb\u7edf\u7f13\u5b58<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">sync<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token number\">3<\/span> <span class=\"token operator\">|<\/span> <span class=\"token function\">sudo<\/span> <span class=\"token function\">tee<\/span> \/proc\/sys\/vm\/drop_caches<\/p>\n<p><span class=\"token comment\"># 3. \u9650\u5236 Docker \u5185\u5b58\u4f7f\u7528<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">mkdir<\/span> <span class=\"token parameter variable\">-p<\/span> \/etc\/docker<br \/>\n<span class=\"token function\">cat<\/span> <span class=\"token operator\">&lt;&lt;<\/span><span class=\"token string\">EOF<span class=\"token bash punctuation\"> <span class=\"token operator\">|<\/span> <span class=\"token function\">sudo<\/span> <span class=\"token function\">tee<\/span> \/etc\/docker\/daemon.json<\/span><br \/>\n{<br \/>\n  &#034;max-concurrent-downloads&#034;: 3,<br \/>\n  &#034;log-driver&#034;: &#034;json-file&#034;,<br \/>\n  &#034;log-opts&#034;: {<br \/>\n    &#034;max-size&#034;: &#034;10m&#034;,<br \/>\n    &#034;max-file&#034;: &#034;3&#034;<br \/>\n  }<br \/>\n}<br \/>\nEOF<\/span><\/p>\n<p><span class=\"token comment\"># 4. \u91cd\u542f Docker \u670d\u52a1<\/span><br \/>\n<span class=\"token function\">sudo<\/span> systemctl restart <span class=\"token function\">docker<\/span><\/p>\n<p><span class=\"token comment\"># 5. \u6e05\u7406 Docker \u5197\u4f59\u6570\u636e<\/span><br \/>\n<span class=\"token function\">docker<\/span> system prune <span class=\"token parameter variable\">-af<\/span><\/p>\n<p><span class=\"token comment\"># 6. \u5b89\u88c5\u76d1\u63a7\u5de5\u5177<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">apt<\/span> update<br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">apt<\/span> <span class=\"token function\">install<\/span> <span class=\"token parameter variable\">-y<\/span> <span class=\"token function\">htop<\/span> iotop<\/p>\n<p><span class=\"token comment\"># 7. \u8bbe\u7f6e\u6587\u4ef6\u63cf\u8ff0\u7b26\u9650\u5236<\/span><br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#039;* soft nofile 65536&#039;<\/span> <span class=\"token operator\">|<\/span> <span class=\"token function\">sudo<\/span> <span class=\"token function\">tee<\/span> <span class=\"token parameter variable\">-a<\/span> \/etc\/security\/limits.conf<br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#039;* hard nofile 65536&#039;<\/span> <span class=\"token operator\">|<\/span> <span class=\"token function\">sudo<\/span> <span class=\"token function\">tee<\/span> <span class=\"token parameter variable\">-a<\/span> \/etc\/security\/limits.conf<\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;\u2705 \u7cfb\u7edf\u4f18\u5316\u5b8c\u6210&#xff01;&#034;<\/span><\/p>\n<hr \/>\n<h3>\u7b2c 5 \u7ae0 \u5b89\u5168\u7ec4\u4e0e\u7f51\u7edc\u914d\u7f6e<\/h3>\n<h4>5.1 \u4e91\u670d\u52a1\u5668\u5b89\u5168\u7ec4\u914d\u7f6e&#xff08;\u963f\u91cc\u4e91\/\u817e\u8baf\u4e91&#xff09;<\/h4>\n<table>\n<tr>\u7aef\u53e3\u534f\u8bae\u65b9\u5411\u6388\u6743\u5bf9\u8c61\u8bf4\u660e\u4f18\u5148\u7ea7<\/tr>\n<tbody>\n<tr>\n<td>11434<\/td>\n<td>TCP<\/td>\n<td>\u5165\u7ad9<\/td>\n<td>\u5e94\u7528\u670d\u52a1\u5668 IP<\/td>\n<td>Ollama API \u7aef\u53e3<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>19530<\/td>\n<td>TCP<\/td>\n<td>\u5165\u7ad9<\/td>\n<td>\u5e94\u7528\u670d\u52a1\u5668 IP<\/td>\n<td>Milvus \u670d\u52a1\u7aef\u53e3<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>9091<\/td>\n<td>TCP<\/td>\n<td>\u5165\u7ad9<\/td>\n<td>\u5e94\u7528\u670d\u52a1\u5668 IP<\/td>\n<td>Milvus \u7ba1\u7406\u7aef\u53e3<\/td>\n<td>&#x1f7e1; \u4e2d<\/td>\n<\/tr>\n<tr>\n<td>8080<\/td>\n<td>TCP<\/td>\n<td>\u5165\u7ad9<\/td>\n<td>0.0.0.0\/0<\/td>\n<td>Spring Boot \u5e94\u7528<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>22<\/td>\n<td>TCP<\/td>\n<td>\u5165\u7ad9<\/td>\n<td>\u7ba1\u7406 IP<\/td>\n<td>SSH \u8fdc\u7a0b\u767b\u5f55<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>3000<\/td>\n<td>TCP<\/td>\n<td>\u5165\u7ad9<\/td>\n<td>\u7ba1\u7406 IP<\/td>\n<td>Attu Web \u7ba1\u7406<\/td>\n<td>&#x1f7e2; \u4f4e<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>5.2 Linux \u9632\u706b\u5899\u914d\u7f6e&#xff08;UFW&#xff09;<\/h4>\n<p><span class=\"token shebang important\">#!\/bin\/bash<\/span><br \/>\n<span class=\"token comment\"># firewall-setup.sh<\/span><\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;&#x1f680; \u914d\u7f6e\u9632\u706b\u5899&#8230;&#034;<\/span><\/p>\n<p><span class=\"token comment\"># 1. \u5b89\u88c5 UFW&#xff08;\u5982\u679c\u672a\u5b89\u88c5&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">apt<\/span> update<br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">apt<\/span> <span class=\"token function\">install<\/span> <span class=\"token parameter variable\">-y<\/span> ufw<\/p>\n<p><span class=\"token comment\"># 2. \u91cd\u7f6e\u9632\u706b\u5899\u89c4\u5219<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw reset<\/p>\n<p><span class=\"token comment\"># 3. \u8bbe\u7f6e\u9ed8\u8ba4\u7b56\u7565<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw default deny incoming<br \/>\n<span class=\"token function\">sudo<\/span> ufw default allow outgoing<\/p>\n<p><span class=\"token comment\"># 4. \u5141\u8bb8 SSH&#xff08;\u9632\u6b62\u88ab\u9501\u5728\u5916\u9762&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw allow <span class=\"token number\">22<\/span>\/tcp<\/p>\n<p><span class=\"token comment\"># 5. \u5141\u8bb8 Ollama \u7aef\u53e3&#xff08;\u4ec5\u5185\u7f51&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw allow from <span class=\"token number\">192.168<\/span>.1.0\/24 to any port <span class=\"token number\">11434<\/span> proto tcp<\/p>\n<p><span class=\"token comment\"># 6. \u5141\u8bb8 Milvus \u7aef\u53e3&#xff08;\u4ec5\u5185\u7f51&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw allow from <span class=\"token number\">192.168<\/span>.1.0\/24 to any port <span class=\"token number\">19530<\/span> proto tcp<br \/>\n<span class=\"token function\">sudo<\/span> ufw allow from <span class=\"token number\">192.168<\/span>.1.0\/24 to any port <span class=\"token number\">9091<\/span> proto tcp<\/p>\n<p><span class=\"token comment\"># 7. \u5141\u8bb8\u5e94\u7528\u7aef\u53e3&#xff08;\u516c\u7f51&#xff09;<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw allow <span class=\"token number\">8080<\/span>\/tcp<\/p>\n<p><span class=\"token comment\"># 8. \u542f\u7528\u9632\u706b\u5899<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw <span class=\"token builtin class-name\">enable<\/span><\/p>\n<p><span class=\"token comment\"># 9. \u67e5\u770b\u72b6\u6001<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw status verbose<\/p>\n<p><span class=\"token comment\"># 10. \u67e5\u770b\u65e5\u5fd7<\/span><br \/>\n<span class=\"token function\">sudo<\/span> ufw logging on<br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">tail<\/span> <span class=\"token parameter variable\">-f<\/span> \/var\/log\/ufw.log<\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;\u2705 \u9632\u706b\u5899\u914d\u7f6e\u5b8c\u6210&#xff01;&#034;<\/span><\/p>\n<h4>5.3 Nginx \u53cd\u5411\u4ee3\u7406\u914d\u7f6e&#xff08;\u751f\u4ea7\u73af\u5883\u63a8\u8350&#xff09;<\/h4>\n<p># \/etc\/nginx\/sites-available\/ollama<br \/>\nserver {<br \/>\n    listen 80;<br \/>\n    server_name api.yourdomain.com;<\/p>\n<p>    # \u9650\u5236\u8bf7\u6c42\u9891\u7387&#xff08;\u9632\u6b62 DDoS&#xff09;<br \/>\n    limit_req_zone $binary_remote_addr zone&#061;ollama:10m rate&#061;10r\/s;<\/p>\n<p>    location \/ {<br \/>\n        # \u5e94\u7528\u9650\u6d41<br \/>\n        limit_req zone&#061;ollama burst&#061;20 nodelay;<\/p>\n<p>        # \u53cd\u5411\u4ee3\u7406\u5230 Ollama<br \/>\n        proxy_pass http:\/\/127.0.0.1:11434;<br \/>\n        proxy_http_version 1.1;<br \/>\n        proxy_set_header Host $host;<br \/>\n        proxy_set_header X-Real-IP $remote_addr;<br \/>\n        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;<\/p>\n<p>        # SSE \u6d41\u5f0f\u54cd\u5e94\u652f\u6301<br \/>\n        proxy_set_header Connection &#034;&#034;;<br \/>\n        proxy_buffering off;<br \/>\n        proxy_cache off;<br \/>\n        chunked_transfer_encoding off;<\/p>\n<p>        # \u8d85\u65f6\u8bbe\u7f6e<br \/>\n        proxy_connect_timeout 60s;<br \/>\n        proxy_send_timeout 300s;<br \/>\n        proxy_read_timeout 300s;<br \/>\n    }<\/p>\n<p>    # \u5065\u5eb7\u68c0\u67e5\u63a5\u53e3<br \/>\n    location \/health {<br \/>\n        access_log off;<br \/>\n        return 200 &#034;OK&#034;;<br \/>\n        add_header Content-Type text\/plain;<br \/>\n    }<br \/>\n}<\/p>\n<h4>5.4 JWT \u8ba4\u8bc1\u4e2d\u95f4\u4ef6&#xff08;Spring Security&#xff09;<\/h4>\n<p><span class=\"token comment\">\/\/ OllamaAuthFilter.java<\/span><br \/>\n<span class=\"token annotation punctuation\">&#064;Component<\/span><br \/>\n<span class=\"token keyword\">public<\/span> <span class=\"token keyword\">class<\/span> <span class=\"token class-name\">OllamaAuthFilter<\/span> <span class=\"token keyword\">extends<\/span> <span class=\"token class-name\">OncePerRequestFilter<\/span> <span class=\"token punctuation\">{<\/span><\/p>\n<p>    <span class=\"token annotation punctuation\">&#064;Autowired<\/span><br \/>\n    <span class=\"token keyword\">private<\/span> <span class=\"token class-name\">JwtTokenProvider<\/span> jwtTokenProvider<span class=\"token punctuation\">;<\/span><\/p>\n<p>    <span class=\"token annotation punctuation\">&#064;Override<\/span><br \/>\n    <span class=\"token keyword\">protected<\/span> <span class=\"token keyword\">void<\/span> <span class=\"token function\">doFilterInternal<\/span><span class=\"token punctuation\">(<\/span><span class=\"token class-name\">HttpServletRequest<\/span> request<span class=\"token punctuation\">,<\/span><br \/>\n                                    <span class=\"token class-name\">HttpServletResponse<\/span> response<span class=\"token punctuation\">,<\/span><br \/>\n                                    <span class=\"token class-name\">FilterChain<\/span> filterChain<span class=\"token punctuation\">)<\/span> <span class=\"token punctuation\">{<\/span><\/p>\n<p>        <span class=\"token comment\">\/\/ 1. \u83b7\u53d6 Token<\/span><br \/>\n        <span class=\"token class-name\">String<\/span> token <span class=\"token operator\">&#061;<\/span> <span class=\"token function\">extractToken<\/span><span class=\"token punctuation\">(<\/span>request<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><\/p>\n<p>        <span class=\"token comment\">\/\/ 2. \u9a8c\u8bc1 Token<\/span><br \/>\n        <span class=\"token keyword\">if<\/span> <span class=\"token punctuation\">(<\/span>token <span class=\"token operator\">!&#061;<\/span> <span class=\"token keyword\">null<\/span> <span class=\"token operator\">&amp;&amp;<\/span> jwtTokenProvider<span class=\"token punctuation\">.<\/span><span class=\"token function\">validateToken<\/span><span class=\"token punctuation\">(<\/span>token<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token punctuation\">{<\/span><br \/>\n            <span class=\"token comment\">\/\/ 3. \u8bbe\u7f6e\u7528\u6237\u4e0a\u4e0b\u6587<\/span><br \/>\n            <span class=\"token class-name\">UsernamePasswordAuthenticationToken<\/span> auth <span class=\"token operator\">&#061;<\/span><br \/>\n                jwtTokenProvider<span class=\"token punctuation\">.<\/span><span class=\"token function\">getAuthentication<\/span><span class=\"token punctuation\">(<\/span>token<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><br \/>\n            <span class=\"token class-name\">SecurityContextHolder<\/span><span class=\"token punctuation\">.<\/span><span class=\"token function\">getContext<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span><span class=\"token function\">setAuthentication<\/span><span class=\"token punctuation\">(<\/span>auth<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><\/p>\n<p>            <span class=\"token comment\">\/\/ 4. \u7ee7\u7eed\u8fc7\u6ee4\u94fe<\/span><br \/>\n            <span class=\"token keyword\">try<\/span> <span class=\"token punctuation\">{<\/span><br \/>\n                filterChain<span class=\"token punctuation\">.<\/span><span class=\"token function\">doFilter<\/span><span class=\"token punctuation\">(<\/span>request<span class=\"token punctuation\">,<\/span> response<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><br \/>\n            <span class=\"token punctuation\">}<\/span> <span class=\"token keyword\">catch<\/span> <span class=\"token punctuation\">(<\/span><span class=\"token class-name\">Exception<\/span> e<span class=\"token punctuation\">)<\/span> <span class=\"token punctuation\">{<\/span><br \/>\n                log<span class=\"token punctuation\">.<\/span><span class=\"token function\">error<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\u8fc7\u6ee4\u94fe\u6267\u884c\u5931\u8d25&#034;<\/span><span class=\"token punctuation\">,<\/span> e<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><br \/>\n            <span class=\"token punctuation\">}<\/span><br \/>\n        <span class=\"token punctuation\">}<\/span> <span class=\"token keyword\">else<\/span> <span class=\"token punctuation\">{<\/span><br \/>\n            <span class=\"token comment\">\/\/ 5. \u8fd4\u56de 401<\/span><br \/>\n            response<span class=\"token punctuation\">.<\/span><span class=\"token function\">setStatus<\/span><span class=\"token punctuation\">(<\/span><span class=\"token class-name\">HttpServletResponse<\/span><span class=\"token punctuation\">.<\/span><span class=\"token constant\">SC_UNAUTHORIZED<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><br \/>\n        <span class=\"token punctuation\">}<\/span><br \/>\n    <span class=\"token punctuation\">}<\/span><\/p>\n<p>    <span class=\"token keyword\">private<\/span> <span class=\"token class-name\">String<\/span> <span class=\"token function\">extractToken<\/span><span class=\"token punctuation\">(<\/span><span class=\"token class-name\">HttpServletRequest<\/span> request<span class=\"token punctuation\">)<\/span> <span class=\"token punctuation\">{<\/span><br \/>\n        <span class=\"token class-name\">String<\/span> bearer <span class=\"token operator\">&#061;<\/span> request<span class=\"token punctuation\">.<\/span><span class=\"token function\">getHeader<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;Authorization&#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><br \/>\n        <span class=\"token keyword\">if<\/span> <span class=\"token punctuation\">(<\/span>bearer <span class=\"token operator\">!&#061;<\/span> <span class=\"token keyword\">null<\/span> <span class=\"token operator\">&amp;&amp;<\/span> bearer<span class=\"token punctuation\">.<\/span><span class=\"token function\">startsWith<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;Bearer &#034;<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token punctuation\">{<\/span><br \/>\n            <span class=\"token keyword\">return<\/span> bearer<span class=\"token punctuation\">.<\/span><span class=\"token function\">substring<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">7<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><br \/>\n        <span class=\"token punctuation\">}<\/span><br \/>\n        <span class=\"token keyword\">return<\/span> <span class=\"token keyword\">null<\/span><span class=\"token punctuation\">;<\/span><br \/>\n    <span class=\"token punctuation\">}<\/span><br \/>\n<span class=\"token punctuation\">}<\/span><\/p>\n<hr \/>\n<h3>\u7b2c 6 \u7ae0 \u5e38\u89c1\u95ee\u9898\u6392\u67e5\u6e05\u5355<\/h3>\n<h4>6.1 \u542f\u52a8\u7c7b\u95ee\u9898<\/h4>\n<table>\n<tr>\u95ee\u9898\u75c7\u72b6\u539f\u56e0\u89e3\u51b3\u65b9\u6848\u4f18\u5148\u7ea7<\/tr>\n<tbody>\n<tr>\n<td>\u670d\u52a1\u65e0\u6cd5\u542f\u52a8<\/td>\n<td>\u7aef\u53e3\u5360\u7528<\/td>\n<td>11434 \u5df2\u88ab\u5360\u7528<\/td>\n<td>lsof -i:11434 \u67e5\u627e\u5e76 kill<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>\u6a21\u578b\u52a0\u8f7d\u5931\u8d25<\/td>\n<td>OOM Killed<\/td>\n<td>\u5185\u5b58\u4e0d\u8db3<\/td>\n<td>\u589e\u52a0 Swap \u6216\u6362\u5c0f\u6a21\u578b<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>\u8fde\u63a5\u88ab\u62d2\u7edd<\/td>\n<td>Connection refused<\/td>\n<td>\u9632\u706b\u5899\u963b\u6321<\/td>\n<td>\u68c0\u67e5\u5b89\u5168\u7ec4\/UFW \u914d\u7f6e<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>\u6743\u9650\u9519\u8bef<\/td>\n<td>Permission denied<\/td>\n<td>\u6587\u4ef6\u6743\u9650\u95ee\u9898<\/td>\n<td>chown -R $USER ~\/.ollama<\/td>\n<td>&#x1f7e1; \u4e2d<\/td>\n<\/tr>\n<tr>\n<td>Docker \u542f\u52a8\u5931\u8d25<\/td>\n<td>\u5bb9\u5668\u53cd\u590d\u91cd\u542f<\/td>\n<td>\u8d44\u6e90\u9650\u5236\u8fc7\u4f4e<\/td>\n<td>\u8c03\u6574 docker-compose \u5185\u5b58\u9650\u5236<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>6.2 \u6027\u80fd\u7c7b\u95ee\u9898<\/h4>\n<table>\n<tr>\u95ee\u9898\u75c7\u72b6\u539f\u56e0\u89e3\u51b3\u65b9\u6848\u4f18\u5148\u7ea7<\/tr>\n<tbody>\n<tr>\n<td>\u9996\u5b57\u5ef6\u8fdf&gt;5s<\/td>\n<td>\u54cd\u5e94\u6162<\/td>\n<td>\u6a21\u578b\u672a\u9884\u52a0\u8f7d<\/td>\n<td>\u542f\u52a8\u540e\u5148 warmup<\/td>\n<td>&#x1f7e1; \u4e2d<\/td>\n<\/tr>\n<tr>\n<td>\u63a8\u7406\u901f\u5ea6&lt;10 token\/s<\/td>\n<td>\u751f\u6210\u6162<\/td>\n<td>CPU \u7ebf\u7a0b\u8fc7\u591a<\/td>\n<td>\u9650\u5236 OLLAMA_NUM_THREADS&#061;2<\/td>\n<td>&#x1f7e1; \u4e2d<\/td>\n<\/tr>\n<tr>\n<td>\u5185\u5b58\u6301\u7eed\u589e\u957f<\/td>\n<td>\u5185\u5b58\u6cc4\u6f0f<\/td>\n<td>\u6a21\u578b\u672a\u91ca\u653e<\/td>\n<td>\u8bbe\u7f6e OLLAMA_KEEP_ALIVE&#061;5m<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>CPU \u6301\u7eed 100%<\/td>\n<td>\u7cfb\u7edf\u5361\u987f<\/td>\n<td>\u5e76\u53d1\u8bf7\u6c42\u8fc7\u591a<\/td>\n<td>\u9650\u5236 max_queue&#061;1<\/td>\n<td>&#x1f534; \u9ad8<\/td>\n<\/tr>\n<tr>\n<td>\u78c1\u76d8 IO \u9ad8<\/td>\n<td>\u7cfb\u7edf\u5361\u987f<\/td>\n<td>Swap \u9891\u7e41\u4f7f\u7528<\/td>\n<td>\u589e\u52a0\u7269\u7406\u5185\u5b58\u6216\u51cf\u5c11\u6a21\u578b<\/td>\n<td>&#x1f7e1; \u4e2d<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h4>6.3 \u6392\u67e5\u547d\u4ee4\u901f\u67e5<\/h4>\n<p><span class=\"token comment\"># 1. \u68c0\u67e5 Ollama \u670d\u52a1\u72b6\u6001<\/span><br \/>\nsystemctl status ollama<\/p>\n<p><span class=\"token comment\"># 2. \u67e5\u770b Ollama \u65e5\u5fd7<\/span><br \/>\njournalctl <span class=\"token parameter variable\">-u<\/span> ollama <span class=\"token parameter variable\">-f<\/span><\/p>\n<p><span class=\"token comment\"># 3. \u68c0\u67e5\u7aef\u53e3\u5360\u7528<\/span><br \/>\n<span class=\"token function\">lsof<\/span> <span class=\"token parameter variable\">-i:11434<\/span><br \/>\n<span class=\"token function\">netstat<\/span> <span class=\"token parameter variable\">-tlnp<\/span> <span class=\"token operator\">|<\/span> <span class=\"token function\">grep<\/span> <span class=\"token number\">11434<\/span><\/p>\n<p><span class=\"token comment\"># 4. \u67e5\u770b\u6a21\u578b\u5217\u8868<\/span><br \/>\nollama list<\/p>\n<p><span class=\"token comment\"># 5. \u67e5\u770b\u8fd0\u884c\u4e2d\u7684\u6a21\u578b<\/span><br \/>\nollama <span class=\"token function\">ps<\/span><\/p>\n<p><span class=\"token comment\"># 6. \u76d1\u63a7\u7cfb\u7edf\u8d44\u6e90<\/span><br \/>\n<span class=\"token function\">htop<\/span><br \/>\n<span class=\"token function\">free<\/span> <span class=\"token parameter variable\">-h<\/span><br \/>\n<span class=\"token function\">df<\/span> <span class=\"token parameter variable\">-h<\/span><\/p>\n<p><span class=\"token comment\"># 7. \u67e5\u770b Docker \u5bb9\u5668\u72b6\u6001<\/span><br \/>\n<span class=\"token function\">docker<\/span> <span class=\"token function\">ps<\/span><br \/>\n<span class=\"token function\">docker<\/span> stats ollama<\/p>\n<p><span class=\"token comment\"># 8. \u6d4b\u8bd5 API \u8fde\u63a5<\/span><br \/>\n<span class=\"token function\">curl<\/span> http:\/\/localhost:11434\/api\/tags<\/p>\n<p><span class=\"token comment\"># 9. \u67e5\u770b\u6a21\u578b\u8be6\u7ec6\u4fe1\u606f<\/span><br \/>\nollama show <span class=\"token operator\">&lt;<\/span>model-name<span class=\"token operator\">&gt;<\/span><\/p>\n<p><span class=\"token comment\"># 10. \u6e05\u7406\u672a\u4f7f\u7528\u6a21\u578b<\/span><br \/>\nollama prune<\/p>\n<hr \/>\n<h3>\u7b2c 7 \u7ae0 \u751f\u4ea7\u73af\u5883\u90e8\u7f72\u6700\u4f73\u5b9e\u8df5<\/h3>\n<h4>7.1 Docker Compose \u751f\u4ea7\u6a21\u677f<\/h4>\n<p><span class=\"token comment\"># docker-compose.prod.yml<\/span><br \/>\n<span class=\"token key atrule\">services<\/span><span class=\"token punctuation\">:<\/span><br \/>\n  <span class=\"token key atrule\">ollama<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token key atrule\">container_name<\/span><span class=\"token punctuation\">:<\/span> ollama<br \/>\n    <span class=\"token key atrule\">image<\/span><span class=\"token punctuation\">:<\/span> ollama\/ollama<span class=\"token punctuation\">:<\/span>latest<br \/>\n    <span class=\"token key atrule\">ports<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token string\">&#034;11434:11434&#034;<\/span><br \/>\n    <span class=\"token key atrule\">volumes<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> .\/ollama<span class=\"token punctuation\">&#8211;<\/span>data<span class=\"token punctuation\">:<\/span>\/root\/.ollama<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> .\/models<span class=\"token punctuation\">:<\/span>\/models<br \/>\n    <span class=\"token key atrule\">environment<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_HOST&#061;0.0.0.0<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_KEEP_ALIVE&#061;5m<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_NUM_THREADS&#061;2<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> OLLAMA_MAX_LOADED_MODELS&#061;1<br \/>\n    <span class=\"token key atrule\">deploy<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token key atrule\">resources<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token key atrule\">limits<\/span><span class=\"token punctuation\">:<\/span><br \/>\n          <span class=\"token key atrule\">memory<\/span><span class=\"token punctuation\">:<\/span> 2G<br \/>\n          <span class=\"token key atrule\">cpus<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#039;2.0&#039;<\/span><br \/>\n    <span class=\"token key atrule\">restart<\/span><span class=\"token punctuation\">:<\/span> unless<span class=\"token punctuation\">&#8211;<\/span>stopped<br \/>\n    <span class=\"token key atrule\">healthcheck<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token key atrule\">test<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string\">&#034;CMD&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;curl&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;-f&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;http:\/\/localhost:11434\/api\/tags&#034;<\/span><span class=\"token punctuation\">]<\/span><br \/>\n      <span class=\"token key atrule\">interval<\/span><span class=\"token punctuation\">:<\/span> 30s<br \/>\n      <span class=\"token key atrule\">timeout<\/span><span class=\"token punctuation\">:<\/span> 10s<br \/>\n      <span class=\"token key atrule\">retries<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token number\">3<\/span><br \/>\n    <span class=\"token key atrule\">networks<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> rag<span class=\"token punctuation\">&#8211;<\/span>network<br \/>\n    <span class=\"token key atrule\">logging<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token key atrule\">driver<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#034;json-file&#034;<\/span><br \/>\n      <span class=\"token key atrule\">options<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token key atrule\">max-size<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#034;10m&#034;<\/span><br \/>\n        <span class=\"token key atrule\">max-file<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#034;3&#034;<\/span><\/p>\n<p>  <span class=\"token key atrule\">nginx<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token key atrule\">container_name<\/span><span class=\"token punctuation\">:<\/span> nginx<span class=\"token punctuation\">&#8211;<\/span>proxy<br \/>\n    <span class=\"token key atrule\">image<\/span><span class=\"token punctuation\">:<\/span> nginx<span class=\"token punctuation\">:<\/span>alpine<br \/>\n    <span class=\"token key atrule\">ports<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token string\">&#034;80:80&#034;<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token string\">&#034;443:443&#034;<\/span><br \/>\n    <span class=\"token key atrule\">volumes<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> .\/nginx.conf<span class=\"token punctuation\">:<\/span>\/etc\/nginx\/nginx.conf<br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> .\/ssl<span class=\"token punctuation\">:<\/span>\/etc\/nginx\/ssl<br \/>\n    <span class=\"token key atrule\">depends_on<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> ollama<br \/>\n    <span class=\"token key atrule\">restart<\/span><span class=\"token punctuation\">:<\/span> unless<span class=\"token punctuation\">&#8211;<\/span>stopped<br \/>\n    <span class=\"token key atrule\">networks<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> rag<span class=\"token punctuation\">&#8211;<\/span>network<\/p>\n<p><span class=\"token key atrule\">networks<\/span><span class=\"token punctuation\">:<\/span><br \/>\n  <span class=\"token key atrule\">rag-network<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token key atrule\">driver<\/span><span class=\"token punctuation\">:<\/span> bridge<\/p>\n<h4>7.2 \u76d1\u63a7\u544a\u8b66\u914d\u7f6e&#xff08;Prometheus&#xff09;<\/h4>\n<p><span class=\"token comment\"># prometheus.yml<\/span><br \/>\n<span class=\"token key atrule\">scrape_configs<\/span><span class=\"token punctuation\">:<\/span><br \/>\n  <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token key atrule\">job_name<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#039;ollama&#039;<\/span><br \/>\n    <span class=\"token key atrule\">static_configs<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token key atrule\">targets<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string\">&#039;ollama:11434&#039;<\/span><span class=\"token punctuation\">]<\/span><br \/>\n    <span class=\"token key atrule\">metrics_path<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#039;\/api\/metrics&#039;<\/span><br \/>\n    <span class=\"token key atrule\">scrape_interval<\/span><span class=\"token punctuation\">:<\/span> 15s<\/p>\n<p>  <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token key atrule\">job_name<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#039;milvus&#039;<\/span><br \/>\n    <span class=\"token key atrule\">static_configs<\/span><span class=\"token punctuation\">:<\/span><br \/>\n      <span class=\"token punctuation\">&#8211;<\/span> <span class=\"token key atrule\">targets<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token string\">&#039;milvus:9091&#039;<\/span><span class=\"token punctuation\">]<\/span><br \/>\n    <span class=\"token key atrule\">scrape_interval<\/span><span class=\"token punctuation\">:<\/span> 15s<\/p>\n<h4>7.3 \u5907\u4efd\u4e0e\u6062\u590d\u7b56\u7565<\/h4>\n<p><span class=\"token shebang important\">#!\/bin\/bash<\/span><br \/>\n<span class=\"token comment\"># backup-ollama.sh<\/span><\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;&#x1f680; \u5f00\u59cb\u5907\u4efd Ollama \u6570\u636e&#8230;&#034;<\/span><\/p>\n<p><span class=\"token comment\"># 1. \u521b\u5efa\u5907\u4efd\u76ee\u5f55<\/span><br \/>\n<span class=\"token assign-left variable\">BACKUP_DIR<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#034;\/backup\/ollama\/<span class=\"token variable\"><span class=\"token variable\">$(<\/span><span class=\"token function\">date<\/span> &#043;%Y%m%d_%H%M%S<span class=\"token variable\">)<\/span><\/span>&#034;<\/span><br \/>\n<span class=\"token function\">mkdir<\/span> <span class=\"token parameter variable\">-p<\/span> <span class=\"token variable\">$BACKUP_DIR<\/span><\/p>\n<p><span class=\"token comment\"># 2. \u5907\u4efd\u6a21\u578b\u6570\u636e<\/span><br \/>\n<span class=\"token function\">cp<\/span> <span class=\"token parameter variable\">-r<\/span> ~\/.ollama\/models <span class=\"token variable\">$BACKUP_DIR<\/span>\/models<\/p>\n<p><span class=\"token comment\"># 3. \u5907\u4efd\u914d\u7f6e\u6587\u4ef6<\/span><br \/>\n<span class=\"token function\">cp<\/span> ~\/.ollama\/Modelfile* <span class=\"token variable\">$BACKUP_DIR<\/span>\/ <span class=\"token operator\"><span class=\"token file-descriptor important\">2<\/span>&gt;<\/span>\/dev\/null <span class=\"token operator\">||<\/span> <span class=\"token boolean\">true<\/span><\/p>\n<p><span class=\"token comment\"># 4. \u538b\u7f29\u5907\u4efd<\/span><br \/>\n<span class=\"token function\">tar<\/span> <span class=\"token parameter variable\">-czf<\/span> <span class=\"token variable\">$BACKUP_DIR<\/span>.tar.gz <span class=\"token parameter variable\">-C<\/span> <span class=\"token variable\">$BACKUP_DIR<\/span> <span class=\"token builtin class-name\">.<\/span><\/p>\n<p><span class=\"token comment\"># 5. \u6e05\u7406\u4e34\u65f6\u76ee\u5f55<\/span><br \/>\n<span class=\"token function\">rm<\/span> <span class=\"token parameter variable\">-rf<\/span> <span class=\"token variable\">$BACKUP_DIR<\/span><\/p>\n<p><span class=\"token comment\"># 6. \u6e05\u7406\u65e7\u5907\u4efd&#xff08;\u4fdd\u7559 7 \u5929&#xff09;<\/span><br \/>\n<span class=\"token function\">find<\/span> \/backup\/ollama <span class=\"token parameter variable\">-name<\/span> <span class=\"token string\">&#034;*.tar.gz&#034;<\/span> <span class=\"token parameter variable\">-mtime<\/span> &#043;7 <span class=\"token parameter variable\">-delete<\/span><\/p>\n<p><span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#034;\u2705 \u5907\u4efd\u5b8c\u6210&#xff1a;<span class=\"token variable\">$BACKUP_DIR<\/span>.tar.gz&#034;<\/span><\/p>\n<p><span class=\"token comment\"># \u6062\u590d\u547d\u4ee4<\/span><br \/>\n<span class=\"token comment\"># tar -xzf backup.tar.gz -C ~\/.ollama\/<\/span><\/p>\n<h4>7.4 \u6027\u80fd\u57fa\u51c6\u6d4b\u8bd5<\/h4>\n<p><span class=\"token comment\"># benchmark_ollama.py<\/span><br \/>\n<span class=\"token keyword\">import<\/span> requests<br \/>\n<span class=\"token keyword\">import<\/span> time<br \/>\n<span class=\"token keyword\">import<\/span> statistics<\/p>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">benchmark_model<\/span><span class=\"token punctuation\">(<\/span>model_name<span class=\"token punctuation\">:<\/span> <span class=\"token builtin\">str<\/span><span class=\"token punctuation\">,<\/span> num_requests<span class=\"token punctuation\">:<\/span> <span class=\"token builtin\">int<\/span> <span class=\"token operator\">&#061;<\/span> <span class=\"token number\">10<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token triple-quoted-string string\">&#034;&#034;&#034;\u6027\u80fd\u57fa\u51c6\u6d4b\u8bd5&#034;&#034;&#034;<\/span><\/p>\n<p>    latencies <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">]<\/span><br \/>\n    throughputs <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">]<\/span><\/p>\n<p>    <span class=\"token keyword\">for<\/span> i <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span>num_requests<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        start <span class=\"token operator\">&#061;<\/span> time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>        response <span class=\"token operator\">&#061;<\/span> requests<span class=\"token punctuation\">.<\/span>post<span class=\"token punctuation\">(<\/span><br \/>\n            <span class=\"token string\">&#034;http:\/\/localhost:11434\/api\/generate&#034;<\/span><span class=\"token punctuation\">,<\/span><br \/>\n            json<span class=\"token operator\">&#061;<\/span><span class=\"token punctuation\">{<\/span><br \/>\n                <span class=\"token string\">&#034;model&#034;<\/span><span class=\"token punctuation\">:<\/span> model_name<span class=\"token punctuation\">,<\/span><br \/>\n                <span class=\"token string\">&#034;prompt&#034;<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token string\">&#034;\u8bf7\u7b80\u8981\u4ecb\u7ecd RAG \u6280\u672f&#xff0c;\u7ea6 100 \u5b57&#034;<\/span><span class=\"token punctuation\">,<\/span><br \/>\n                <span class=\"token string\">&#034;stream&#034;<\/span><span class=\"token punctuation\">:<\/span> <span class=\"token boolean\">False<\/span><br \/>\n            <span class=\"token punctuation\">}<\/span><br \/>\n        <span class=\"token punctuation\">)<\/span><\/p>\n<p>        end <span class=\"token operator\">&#061;<\/span> time<span class=\"token punctuation\">.<\/span>time<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        latency <span class=\"token operator\">&#061;<\/span> end <span class=\"token operator\">&#8211;<\/span> start<\/p>\n<p>        <span class=\"token comment\"># \u8ba1\u7b97\u541e\u5410\u91cf<\/span><br \/>\n        token_count <span class=\"token operator\">&#061;<\/span> response<span class=\"token punctuation\">.<\/span>json<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>get<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;eval_count&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">0<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        throughput <span class=\"token operator\">&#061;<\/span> token_count <span class=\"token operator\">\/<\/span> latency<\/p>\n<p>        latencies<span class=\"token punctuation\">.<\/span>append<span class=\"token punctuation\">(<\/span>latency<span class=\"token punctuation\">)<\/span><br \/>\n        throughputs<span class=\"token punctuation\">.<\/span>append<span class=\"token punctuation\">(<\/span>throughput<span class=\"token punctuation\">)<\/span><\/p>\n<p>        <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;\u8bf7\u6c42 <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>i<span class=\"token operator\">&#043;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">: \u5ef6\u8fdf&#061;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>latency<span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.2f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">s, \u541e\u5410\u91cf&#061;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>throughput<span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.1f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\"> token\/s&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>    <span class=\"token comment\"># \u7edf\u8ba1\u7ed3\u679c<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;\\\\n&#034;<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token string\">&#034;&#061;&#034;<\/span><span class=\"token operator\">*<\/span><span class=\"token number\">50<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;\u6a21\u578b&#xff1a;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>model_name<span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;\u5e73\u5747\u5ef6\u8fdf&#xff1a;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>statistics<span class=\"token punctuation\">.<\/span>mean<span class=\"token punctuation\">(<\/span>latencies<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.2f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">s&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;P95 \u5ef6\u8fdf&#xff1a;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span><span class=\"token builtin\">sorted<\/span><span class=\"token punctuation\">(<\/span>latencies<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">[<\/span><span class=\"token builtin\">int<\/span><span class=\"token punctuation\">(<\/span>num_requests<span class=\"token operator\">*<\/span><span class=\"token number\">0.95<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.2f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">s&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;\u5e73\u5747\u541e\u5410\u91cf&#xff1a;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>statistics<span class=\"token punctuation\">.<\/span>mean<span class=\"token punctuation\">(<\/span>throughputs<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.1f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\"> token\/s&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#034;\u6700\u5c0f\u541e\u5410\u91cf&#xff1a;<\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span><span class=\"token builtin\">min<\/span><span class=\"token punctuation\">(<\/span>throughputs<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.1f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\"> token\/s&#034;<\/span><\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;&#061;&#034;<\/span><span class=\"token operator\">*<\/span><span class=\"token number\">50<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u8fd0\u884c\u6d4b\u8bd5<\/span><br \/>\nbenchmark_model<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;qwen2.5:0.5b&#034;<\/span><span class=\"token punctuation\">,<\/span> num_requests<span class=\"token operator\">&#061;<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<hr \/>\n<p>\u672c\u4e13\u9898\u5b8c<\/p>\n<p>&#x1f4cc; \u6838\u5fc3\u8981\u70b9\u603b\u7ed3&#xff1a;<\/p>\n<li>Ollama \u662f\u672c\u5730\u90e8\u7f72 LLM \u7684\u6700\u4f73\u9009\u62e9&#xff0c;\u4e00\u952e\u62c9\u53d6\u6a21\u578b<\/li>\n<li>Q4 \u91cf\u5316\u53ef\u5728\u7cbe\u5ea6\u635f\u5931&lt;5% \u7684\u60c5\u51b5\u4e0b\u964d\u4f4e 70% \u5185\u5b58\u5360\u7528<\/li>\n<li>2C4G \u670d\u52a1\u5668\u5fc5\u987b\u914d\u7f6e 4G&#043; Swap \u865a\u62df\u5185\u5b58<\/li>\n<li>OLLAMA_KEEP_ALIVE&#061;5m \u9632\u6b62\u5185\u5b58\u6cc4\u6f0f<\/li>\n<li>\u4e91\u670d\u52a1\u5668\u5b89\u5168\u7ec4\u4ec5\u5f00\u653e\u5fc5\u8981\u7aef\u53e3&#xff0c;\u5185\u7f51\u901a\u4fe1<\/li>\n<li>Nginx \u53cd\u5411\u4ee3\u7406 &#043; JWT \u8ba4\u8bc1\u662f\u751f\u4ea7\u73af\u5883\u6807\u914d<\/li>\n<li>\u5b9a\u671f\u5907\u4efd\u6a21\u578b\u6570\u636e&#xff0c;\u9632\u6b62\u610f\u5916\u4e22\u5931<\/li>\n<p>&#x1f4d6; \u4e0b\u4e13\u9898\u9884\u544a&#xff1a;\u300aSpring AI \u540e\u7aef\u5f00\u53d1\u5b9e\u6218&#xff1a;30 \u884c\u4ee3\u7801\u5b9e\u73b0 RAG \u6838\u5fc3\u903b\u8f91\u300b\u2014\u2014 \u8be6\u89e3 Spring AI \u96c6\u6210\u3001\u5411\u91cf\u5b58\u50a8\u670d\u52a1\u3001SSE \u6d41\u5f0f\u54cd\u5e94\u3001\u4f9d\u8d56\u51b2\u7a81\u89e3\u51b3<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u4e13\u9898\u56db&#xff1a;\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18&#xff1a;\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b<br \/>\n&#x1f4da; \u4f01\u4e1a\u7ea7 RAG \u667a\u80fd\u95ee\u7b54\u7cfb\u7edf\u5168\u6808\u5b9e\u65bd\u6307\u5357<br \/>\n\u7b2c\u56db\u90e8\u5206&#xff1a;\u5927\u6a21\u578b\u672c\u5730\u90e8\u7f72\u4e0e\u4f18\u5316\u7b2c 1 \u7ae0 Ollama \u6982\u8ff0\u4e0e\u6838\u5fc3\u4f18\u52bf<br \/>\n1.1 \u4ec0\u4e48\u662f Ollama&#xff1f;<br \/>\nOllama \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u672c\u5730\u5927\u8bed\u8a00\u6a21\u578b\u8fd0\u884c\u5e73\u53f0&#xff0c;\u4e13\u4e3a\u7b80\u5316 LLM \u90e8\u7f72\u800c\u8bbe\u8ba1\u3002\u5b83\u8ba9\u5f00\u53d1\u8005\u80fd\u591f\u5728\u672c\u5730\u670d\u52a1\u5668\u4e0a\u8f7b\u677e\u8fd0\u884c\u3001\u7ba1\u7406\u548c\u4f18\u5316\u5927\u8bed\u8a00\u6a21\u578b\u3002<br \/>\n\u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[101,50,371],"topic":[],"class_list":["post-86009","post","type-post","status-publish","format-standard","hentry","category-server","tag-spring","tag-50","tag-371"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v20.3 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/86009.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u4e13\u9898\u56db&#xff1a;\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18&#xff1a;\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b &#x1f4da; \u4f01\u4e1a\u7ea7 RAG \u667a\u80fd\u95ee\u7b54\u7cfb\u7edf\u5168\u6808\u5b9e\u65bd\u6307\u5357 \u7b2c\u56db\u90e8\u5206&#xff1a;\u5927\u6a21\u578b\u672c\u5730\u90e8\u7f72\u4e0e\u4f18\u5316\u7b2c 1 \u7ae0 Ollama \u6982\u8ff0\u4e0e\u6838\u5fc3\u4f18\u52bf 1.1 \u4ec0\u4e48\u662f Ollama&#xff1f; Ollama \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u672c\u5730\u5927\u8bed\u8a00\u6a21\u578b\u8fd0\u884c\u5e73\u53f0&#xff0c;\u4e13\u4e3a\u7b80\u5316 LLM \u90e8\u7f72\u800c\u8bbe\u8ba1\u3002\u5b83\u8ba9\u5f00\u53d1\u8005\u80fd\u591f\u5728\u672c\u5730\u670d\u52a1\u5668\u4e0a\u8f7b\u677e\u8fd0\u884c\u3001\u7ba1\u7406\u548c\u4f18\u5316\u5927\u8bed\u8a00\u6a21\u578b\u3002 \u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/86009.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-07-27T18:16:58+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"8 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/86009.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/86009.html\",\"name\":\"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-07-27T18:16:58+00:00\",\"dateModified\":\"2026-07-27T18:16:58+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/86009.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/86009.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/86009.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/86009.html","og_locale":"zh_CN","og_type":"article","og_title":"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u4e13\u9898\u56db&#xff1a;\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18&#xff1a;\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b &#x1f4da; \u4f01\u4e1a\u7ea7 RAG \u667a\u80fd\u95ee\u7b54\u7cfb\u7edf\u5168\u6808\u5b9e\u65bd\u6307\u5357 \u7b2c\u56db\u90e8\u5206&#xff1a;\u5927\u6a21\u578b\u672c\u5730\u90e8\u7f72\u4e0e\u4f18\u5316\u7b2c 1 \u7ae0 Ollama \u6982\u8ff0\u4e0e\u6838\u5fc3\u4f18\u52bf 1.1 \u4ec0\u4e48\u662f Ollama&#xff1f; Ollama \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u672c\u5730\u5927\u8bed\u8a00\u6a21\u578b\u8fd0\u884c\u5e73\u53f0&#xff0c;\u4e13\u4e3a\u7b80\u5316 LLM \u90e8\u7f72\u800c\u8bbe\u8ba1\u3002\u5b83\u8ba9\u5f00\u53d1\u8005\u80fd\u591f\u5728\u672c\u5730\u670d\u52a1\u5668\u4e0a\u8f7b\u677e\u8fd0\u884c\u3001\u7ba1\u7406\u548c\u4f18\u5316\u5927\u8bed\u8a00\u6a21\u578b\u3002 \u250c\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500","og_url":"https:\/\/www.wsisp.com\/helps\/86009.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-07-27T18:16:58+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"8 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/86009.html","url":"https:\/\/www.wsisp.com\/helps\/86009.html","name":"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-07-27T18:16:58+00:00","dateModified":"2026-07-27T18:16:58+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/86009.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/86009.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/86009.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"\u300aSpring AI + \u5927\u6a21\u578b\u5168\u6808\u5b9e\u6218\u300b\u5b66\u4e60\u624b\u518c\u7cfb\u5217 \u00b7 \u4e13\u9898\u56db\uff1a\u300aOllama \u6a21\u578b\u7ba1\u7406\u4e0e\u8c03\u4f18\uff1a\u8ba9 AI \u6a21\u578b\u5728\u4f4e\u914d\u670d\u52a1\u5668\u4e0a\u6d41\u7545\u8fd0\u884c\u300b"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/86009","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=86009"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/86009\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=86009"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=86009"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=86009"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=86009"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}