-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsearch.json
1 lines (1 loc) · 156 KB
/
search.json
1
[{"title":"python 多线程与多进程","url":"https://www.holmeyoung.com/blog/python-threading-multiprocessing/","content":"<h2 id=\"多进程\"><a href=\"#多进程\" class=\"headerlink\" title=\"多进程\"></a>多进程</h2><h3 id=\"进程池阻塞与非阻塞\"><a href=\"#进程池阻塞与非阻塞\" class=\"headerlink\" title=\"进程池阻塞与非阻塞\"></a>进程池阻塞与非阻塞</h3><p><span id=\"jump\"></span></p>\n<h4 id=\"异步进程池（非阻塞）\"><a href=\"#异步进程池（非阻塞）\" class=\"headerlink\" title=\"异步进程池（非阻塞）\"></a>异步进程池（非阻塞）</h4><figure class=\"highlight python\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"keyword\">from</span> multiprocessing <span class=\"keyword\">import</span> Pool</span><br><span class=\"line\"><span class=\"function\"><span class=\"keyword\">def</span> <span class=\"title\">test</span><span class=\"params\">(i)</span>:</span></span><br><span class=\"line\"> <span class=\"keyword\">print</span> (i)</span><br><span class=\"line\"><span class=\"keyword\">if</span> __name__==<span class=\"string\">\"__main__\"</span>:</span><br><span class=\"line\">\tpool = Pool(processes=<span class=\"number\">10</span>)</span><br><span class=\"line\">\t<span class=\"keyword\">for</span> i <span class=\"keyword\">in</span> xrange(<span class=\"number\">500</span>):</span><br><span class=\"line\">\t\t<span class=\"string\">'''</span></span><br><span class=\"line\"><span class=\"string\">\t\tFor循环中执行步骤：</span></span><br><span class=\"line\"><span class=\"string\">\t\t（1）循环遍历，将500个子进程添加到进程池（相对父进程会阻塞）</span></span><br><span class=\"line\"><span class=\"string\">\t\t（2）每次执行10个子进程，等一个子进程执行完后，立马启动新的子进程。（相对父进程不阻塞）</span></span><br><span class=\"line\"><span class=\"string\">\t\t</span></span><br><span class=\"line\"><span class=\"string\">\t\tapply_async为异步进程池写法。</span></span><br><span class=\"line\"><span class=\"string\">\t\t异步指的是启动子进程的过程，与父进程本身的执行（print）是异步的，而For循环中往进程池添加子进程的过程，与父进程本身的执行却是同步的。</span></span><br><span class=\"line\"><span class=\"string\">\t\t'''</span></span><br><span class=\"line\">\t pool.apply_async(test, args=(i,)) <span class=\"comment\">#维持执行的进程总数为10，当一个进程执行完后启动一个新进程. </span></span><br><span class=\"line\">\t<span class=\"keyword\">print</span> (<span class=\"string\">'test'</span>)</span><br><span class=\"line\">\tpool.close()</span><br><span class=\"line\">\tpool.join()</span><br></pre></td></tr></table></figure>\n<p>执行顺序：For循环内执行了2个步骤，第一步：将500个对象放入进程池（阻塞）。第二步：同时执行10个子进程（非阻塞），有结束的就立即添加，维持10个子进程运行。（apply_async方法的会在执行完for循环的添加步骤后，直接执行后面的print语句，而apply方法会等所有进程池中的子进程运行完以后再执行后面的print语句）</p>\n<p>注意：调用join之前，先调用close或者terminate方法，否则会出错。执行完close后不会有新的进程加入到pool,join函数等待所有子进程结束。</p>\n<h4 id=\"同步进程池（阻塞）\"><a href=\"#同步进程池（阻塞）\" class=\"headerlink\" title=\"同步进程池（阻塞）\"></a>同步进程池（阻塞）</h4><figure class=\"highlight python\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"keyword\">from</span> multiprocessing <span class=\"keyword\">import</span> Pool</span><br><span class=\"line\"><span class=\"function\"><span class=\"keyword\">def</span> <span class=\"title\">test</span><span class=\"params\">(p)</span>:</span></span><br><span class=\"line\"> <span class=\"keyword\">print</span> (p)</span><br><span class=\"line\"> time.sleep(<span class=\"number\">3</span>)</span><br><span class=\"line\"><span class=\"keyword\">if</span> __name__==<span class=\"string\">\"__main__\"</span>:</span><br><span class=\"line\">\tpool = Pool(processes=<span class=\"number\">10</span>)</span><br><span class=\"line\">\t<span class=\"keyword\">for</span> i <span class=\"keyword\">in</span> xrange(<span class=\"number\">500</span>):</span><br><span class=\"line\">\t<span class=\"string\">'''</span></span><br><span class=\"line\"><span class=\"string\">\t实际测试发现，for循环内部执行步骤：</span></span><br><span class=\"line\"><span class=\"string\">\t（1）遍历500个可迭代对象，往进程池放一个子进程</span></span><br><span class=\"line\"><span class=\"string\">\t（2）执行这个子进程，等子进程执行完毕，再往进程池放一个子进程，再执行。（同时只执行一个子进程）</span></span><br><span class=\"line\"><span class=\"string\">\tfor循环执行完毕，再执行print函数。</span></span><br><span class=\"line\"><span class=\"string\">\t'''</span></span><br><span class=\"line\">\t pool.apply(test, args=(i,)) <span class=\"comment\">#维持执行的进程总数为10，当一个进程执行完后启动一个新进程.</span></span><br><span class=\"line\">\t<span class=\"keyword\">print</span> (<span class=\"string\">'test'</span>)</span><br><span class=\"line\">\tpool.close()</span><br><span class=\"line\">\tpool.join()</span><br></pre></td></tr></table></figure>\n<p>说明：for循环内执行的步骤顺序，往进程池中添加一个子进程，执行子进程，等待执行完毕再添加一个子进程…..等500个子进程都执行完了，再执行print (‘test’)。（从结果来看，并没有多进程并发）</p>\n<h4 id=\"REF\"><a href=\"#REF\" class=\"headerlink\" title=\"REF\"></a>REF</h4><blockquote>\n<p><a href=\"https://thief.one/2016/11/24/Multiprocessing-Pool/\" target=\"_blank\" rel=\"noopener\">https://thief.one/2016/11/24/Multiprocessing-Pool/</a></p>\n</blockquote>\n<h3 id=\"多进程pool\"><a href=\"#多进程pool\" class=\"headerlink\" title=\"多进程pool\"></a>多进程pool</h3><p>见 <a href=\"#jump\">多进程.进程池阻塞与非阻塞</a></p>\n<h3 id=\"多进程pool-queue\"><a href=\"#多进程pool-queue\" class=\"headerlink\" title=\"多进程pool queue\"></a>多进程pool queue</h3><figure class=\"highlight python\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"keyword\">import</span> os</span><br><span class=\"line\"><span class=\"keyword\">from</span> multiprocessing <span class=\"keyword\">import</span> Manager, Pool</span><br><span class=\"line\"><span class=\"keyword\">import</span> sys</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"function\"><span class=\"keyword\">def</span> <span class=\"title\">get_hash_queue</span><span class=\"params\">(i, n, image_path, q_put)</span>:</span></span><br><span class=\"line\"> <span class=\"comment\"># print the doing process</span></span><br><span class=\"line\"> sys.stdout.write(<span class=\"string\">\"\\rLoading process: {}/{}\"</span>.format(str(n-i), str(n)))</span><br><span class=\"line\"> sys.stdout.flush()</span><br><span class=\"line\"> <span class=\"string\">'''Dealwith image_path here.'''</span></span><br><span class=\"line\"> q_put.put_nowait([image_path])</span><br><span class=\"line\"> <span class=\"keyword\">return</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"keyword\">if</span> __name__ == <span class=\"string\">\"__main__\"</span>:</span><br><span class=\"line\"> <span class=\"comment\"># init pool and queue</span></span><br><span class=\"line\"> q_get = Manager().Queue()</span><br><span class=\"line\"> q_put = Manager().Queue()</span><br><span class=\"line\"> pool = Pool(processes = <span class=\"number\">4</span>)</span><br><span class=\"line\"></span><br><span class=\"line\"> <span class=\"comment\"># put pic path into get queue: q_get</span></span><br><span class=\"line\"> pics = os.listdir(folder)</span><br><span class=\"line\"> <span class=\"keyword\">for</span> pic <span class=\"keyword\">in</span> pics:</span><br><span class=\"line\"> path = os.path.join(folder, pic)</span><br><span class=\"line\"> q_get.put(path)</span><br><span class=\"line\"> </span><br><span class=\"line\"> <span class=\"comment\"># use multi process to get hash result and put the result into q_put</span></span><br><span class=\"line\"> n = q_get.qsize()</span><br><span class=\"line\"> <span class=\"keyword\">while</span> <span class=\"literal\">True</span>:</span><br><span class=\"line\"> <span class=\"keyword\">try</span>:</span><br><span class=\"line\"> image_path = q_get.get_nowait()</span><br><span class=\"line\"> i = q_get.qsize()</span><br><span class=\"line\"> <span class=\"keyword\">except</span>:</span><br><span class=\"line\"> <span class=\"keyword\">break</span></span><br><span class=\"line\"> <span class=\"keyword\">else</span>:</span><br><span class=\"line\"> pool.apply_async(get_hash_queue, (i, n, image_path, q_put, ))</span><br><span class=\"line\"> pool.close()</span><br><span class=\"line\"> pool.join()</span><br><span class=\"line\"> <span class=\"string\">'''Dealwith q_put here.'''</span></span><br></pre></td></tr></table></figure>\n<h3 id=\"多进程pool-map\"><a href=\"#多进程pool-map\" class=\"headerlink\" title=\"多进程pool.map\"></a>多进程pool.map</h3><figure class=\"highlight python\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"keyword\">import</span> os</span><br><span class=\"line\"><span class=\"keyword\">import</span> cv2</span><br><span class=\"line\"><span class=\"keyword\">import</span> shutil</span><br><span class=\"line\"><span class=\"keyword\">from</span> multiprocessing <span class=\"keyword\">import</span> Pool</span><br><span class=\"line\"><span class=\"keyword\">import</span> argparse</span><br><span class=\"line\"></span><br><span class=\"line\">pic_suffix = set()</span><br><span class=\"line\">pic_suffix.update([<span class=\"string\">'.BMP'</span>, <span class=\"string\">'.DIB'</span>, <span class=\"string\">'.JPEG'</span>, <span class=\"string\">'.JPG'</span>, <span class=\"string\">'.JPE'</span>, <span class=\"string\">'.PNG'</span>, <span class=\"string\">'.PBM'</span>, <span class=\"string\">'.PGM'</span>, <span class=\"string\">'.PPM'</span>, <span class=\"string\">'.SR'</span>, <span class=\"string\">'.RAS'</span>, <span class=\"string\">'.TIFF'</span>, <span class=\"string\">'.TIF'</span>, <span class=\"string\">'.EXR'</span>, <span class=\"string\">'.JP2'</span>])</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"function\"><span class=\"keyword\">def</span> <span class=\"title\">image_judge</span><span class=\"params\">(image_path)</span>:</span></span><br><span class=\"line\"> <span class=\"keyword\">try</span>:</span><br><span class=\"line\"> new_path = os.path.join(<span class=\"string\">'./trash'</span>, os.path.split(image_path)[<span class=\"number\">-1</span>])</span><br><span class=\"line\"> <span class=\"comment\"># not image</span></span><br><span class=\"line\"> <span class=\"keyword\">if</span> os.path.splitext(image_path)[<span class=\"number\">-1</span>].upper() <span class=\"keyword\">not</span> <span class=\"keyword\">in</span> pic_suffix:</span><br><span class=\"line\"> <span class=\"keyword\">print</span> (<span class=\"string\">'Not image %s'</span> % (new_path))</span><br><span class=\"line\"> shutil.move(image_path, <span class=\"string\">'./trash'</span>)</span><br><span class=\"line\"> <span class=\"keyword\">return</span></span><br><span class=\"line\"> </span><br><span class=\"line\"> <span class=\"comment\"># image is none</span></span><br><span class=\"line\"> image = cv2.imread(image_path)</span><br><span class=\"line\"> <span class=\"keyword\">if</span> image <span class=\"keyword\">is</span> <span class=\"literal\">None</span>:</span><br><span class=\"line\"> <span class=\"keyword\">print</span> (<span class=\"string\">'Image none: %s'</span> % (new_path))</span><br><span class=\"line\"> shutil.move(image_path, <span class=\"string\">'./trash'</span>)</span><br><span class=\"line\"> <span class=\"keyword\">return</span></span><br><span class=\"line\"> <span class=\"keyword\">except</span>:</span><br><span class=\"line\"> <span class=\"comment\"># catch except </span></span><br><span class=\"line\"> <span class=\"keyword\">print</span> (<span class=\"string\">'Image error: %s'</span> % (new_path))</span><br><span class=\"line\"> shutil.move(image_path, <span class=\"string\">'./trash'</span>)</span><br><span class=\"line\"> <span class=\"keyword\">return</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"function\"><span class=\"keyword\">def</span> <span class=\"title\">entry_function</span><span class=\"params\">(folder)</span>:</span></span><br><span class=\"line\"> os.makedirs(<span class=\"string\">'./trash'</span>, exist_ok=<span class=\"literal\">True</span>)</span><br><span class=\"line\"> pic_list = []</span><br><span class=\"line\"> pics = os.listdir(folder)</span><br><span class=\"line\"> <span class=\"keyword\">for</span> pic <span class=\"keyword\">in</span> pics:</span><br><span class=\"line\"> pic_list.append(os.path.join(folder, pic))</span><br><span class=\"line\"> pool = Pool()</span><br><span class=\"line\"> pool.map(image_judge, pic_list)</span><br><span class=\"line\"> pool.close()</span><br><span class=\"line\"> pool.join()</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"keyword\">if</span> __name__ == <span class=\"string\">\"__main__\"</span>:</span><br><span class=\"line\"> parser = argparse.ArgumentParser()</span><br><span class=\"line\"> parser.add_argument(<span class=\"string\">'--folder'</span>, type = str, required = <span class=\"literal\">True</span>, help = <span class=\"string\">'path to folder which contains the images'</span>)</span><br><span class=\"line\"> args = parser.parse_args()</span><br><span class=\"line\"></span><br><span class=\"line\"> entry_function(args.folder)</span><br></pre></td></tr></table></figure>\n<h2 id=\"多线程\"><a href=\"#多线程\" class=\"headerlink\" title=\"多线程\"></a>多线程</h2><h3 id=\"多线程queue\"><a href=\"#多线程queue\" class=\"headerlink\" title=\"多线程queue\"></a>多线程queue</h3><figure class=\"highlight python\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"keyword\">import</span> os</span><br><span class=\"line\"><span class=\"keyword\">import</span> threading</span><br><span class=\"line\"><span class=\"keyword\">import</span> queue</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"function\"><span class=\"keyword\">def</span> <span class=\"title\">fetch_img_func</span><span class=\"params\">(q)</span>:</span></span><br><span class=\"line\"> <span class=\"keyword\">while</span> <span class=\"literal\">True</span>:</span><br><span class=\"line\"> <span class=\"keyword\">try</span>:</span><br><span class=\"line\"> <span class=\"comment\"># unblock read from queue</span></span><br><span class=\"line\"> url = q.get_nowait() <span class=\"comment\"># https://example.com/example.jpg</span></span><br><span class=\"line\"> <span class=\"keyword\">except</span>:</span><br><span class=\"line\"> <span class=\"keyword\">break</span></span><br><span class=\"line\"> <span class=\"keyword\">else</span>:</span><br><span class=\"line\"> <span class=\"keyword\">try</span>:</span><br><span class=\"line\"> <span class=\"string\">'''do something with url here'''</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"keyword\">if</span> __name__ == <span class=\"string\">\"__main__\"</span>:</span><br><span class=\"line\"> q = queue.Queue()</span><br><span class=\"line\"></span><br><span class=\"line\"> <span class=\"string\">'''</span></span><br><span class=\"line\"><span class=\"string\"> Put something into queue here.</span></span><br><span class=\"line\"><span class=\"string\"> eg:</span></span><br><span class=\"line\"><span class=\"string\"> url = https://example.com/example.jpg</span></span><br><span class=\"line\"><span class=\"string\"> q.put(url)</span></span><br><span class=\"line\"><span class=\"string\"> '''</span></span><br><span class=\"line\"></span><br><span class=\"line\"> threads = []</span><br><span class=\"line\"> <span class=\"keyword\">for</span> i <span class=\"keyword\">in</span> range(args.thread):</span><br><span class=\"line\"> thread = threading.Thread(target=fetch_img_func, args=(q, ))</span><br><span class=\"line\"> threads.append(thread)</span><br><span class=\"line\"> thread.start()</span><br><span class=\"line\"> <span class=\"keyword\">for</span> i <span class=\"keyword\">in</span> threads:</span><br><span class=\"line\"> i.join()</span><br></pre></td></tr></table></figure>\n<h3 id=\"多线程下载\"><a href=\"#多线程下载\" class=\"headerlink\" title=\"多线程下载\"></a>多线程下载</h3><p><a href=\"https://github.com/Holmeyoung/image-process/blob/master/image_download.py\" target=\"_blank\" rel=\"noopener\">https://github.com/Holmeyoung/image-process/blob/master/image_download.py</a></p>\n","categories":["python"],"tags":["python"]},{"title":"hexo 通过 gitlab-ci 部署到 github","url":"https://www.holmeyoung.com/blog/hexo-gitlab-ci-github/","content":"<h2 id=\"目的\"><a href=\"#目的\" class=\"headerlink\" title=\"目的\"></a>目的</h2><p>将 <code>Markdown</code> 文件提交到 <code>GitLab</code> ，通过 <code>GitLab</code> 的 <code>ci</code> 工具，自动采用 <code>hexo</code> 将其渲染为 <code>html</code> 文件并部署到 <code>GitHub</code> 。可以真正做到</p>\n<blockquote>\n<p>只维护 <code>Markdown</code> 的博客源文件即可</p>\n<p>提交 <code>Markdown</code> 就能看博客</p>\n<p>源文件、配置文件长久备份</p>\n<p>跨平台、跨终端</p>\n</blockquote>\n<h2 id=\"GitHub-操作\"><a href=\"#GitHub-操作\" class=\"headerlink\" title=\"GitHub 操作\"></a>GitHub 操作</h2><h3 id=\"新建-Pages-仓库\"><a href=\"#新建-Pages-仓库\" class=\"headerlink\" title=\"新建 Pages 仓库\"></a>新建 Pages 仓库</h3><p>假如你的 <code>GitHub</code> 用户名为 <code>xxx</code> ，新建 <strong>共有仓库</strong> <code>xxx.github.io</code></p>\n<h3 id=\"获取-tokens\"><a href=\"#获取-tokens\" class=\"headerlink\" title=\"获取 tokens\"></a>获取 tokens</h3><p>在 <code>GitHub</code> 上获取一个 <strong>只用于</strong> 操作 <code>repo</code> 的 <code>Personal access tokens</code></p>\n<p>具体路径为</p>\n<p><code>Settings</code> – <code>Developer settings</code> – <code>Personal access tokens</code> – <code>Generate new token</code></p>\n<blockquote>\n<p>注意权限勾选 <code>repo</code> 即可</p>\n</blockquote>\n<h2 id=\"GitLab-操作\"><a href=\"#GitLab-操作\" class=\"headerlink\" title=\"GitLab 操作\"></a>GitLab 操作</h2><h3 id=\"新建自己的-hexo-仓库\"><a href=\"#新建自己的-hexo-仓库\" class=\"headerlink\" title=\"新建自己的 hexo 仓库\"></a>新建自己的 hexo 仓库</h3><p>在 <code>gitlab</code> 新建一个自己的 <strong>私有仓库</strong> 。在这里我们假设命名为 <code>blog</code> 。</p>\n<h3 id=\"clone一份官方-hexo-仓库到本地\"><a href=\"#clone一份官方-hexo-仓库到本地\" class=\"headerlink\" title=\"clone一份官方 hexo 仓库到本地\"></a>clone一份官方 hexo 仓库到本地</h3><p>地址：<a href=\"https://gitlab.com/pages/hexo\" target=\"_blank\" rel=\"noopener\">https://gitlab.com/pages/hexo</a></p>\n<p>将其 <code>clone</code> 到本地，删除 <code>.git</code> 文件夹，并将该文件夹初始化为自己在 <code>gitlab新建的仓库</code> 即 <code>blog</code> 。</p>\n<h2 id=\"文件配置\"><a href=\"#文件配置\" class=\"headerlink\" title=\"文件配置\"></a>文件配置</h2><ul>\n<li><p>修改 <code>hexo</code> 的 <code>config.yml</code> 文件</p>\n<figure class=\"highlight yaml\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"attr\">url:</span> <span class=\"string\">your-blog-url</span></span><br><span class=\"line\"><span class=\"attr\">root:</span> <span class=\"string\">/</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"attr\">deploy:</span></span><br><span class=\"line\"> <span class=\"attr\">type:</span> <span class=\"string\">git</span></span><br><span class=\"line\"> <span class=\"attr\">repository:</span></span><br><span class=\"line\"> <span class=\"attr\">github:</span> <span class=\"string\">https://personal-access-token:[email protected]/xxx/xxx.github.io.git</span></span><br><span class=\"line\"> <span class=\"attr\">branch:</span> <span class=\"string\">master</span></span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>将 <code>your-blog-url</code> 替换为自己的博客地址</p>\n<p>把上面从 <code>GitHub</code> 获取的 <code>personal-access-token</code> 放到这里来</p>\n<p>将 <code>xxx</code> 替换为自己的用户名</p>\n</blockquote>\n</li>\n</ul>\n<ul>\n<li><p>修改 <code>.gitlab-ci.yml</code> 文件</p>\n<figure class=\"highlight yaml\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"attr\">image:</span> <span class=\"string\">node:10.15.3</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"attr\">cache:</span></span><br><span class=\"line\"> <span class=\"attr\">paths:</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">node_modules/</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"attr\">before_script:</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">git</span> <span class=\"string\">config</span> <span class=\"string\">--global</span> <span class=\"string\">user.name</span> <span class=\"string\">\"github user name\"</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">git</span> <span class=\"string\">config</span> <span class=\"string\">--global</span> <span class=\"string\">user.email</span> <span class=\"string\">\"gitub email\"</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">npm</span> <span class=\"string\">install</span> <span class=\"string\">-g</span> <span class=\"string\">hexo</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">npm</span> <span class=\"string\">install</span> <span class=\"string\">-g</span> <span class=\"string\">hexo-cli</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">npm</span> <span class=\"string\">install</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">npm</span> <span class=\"string\">install</span> <span class=\"string\">hexo-deployer-git</span> <span class=\"string\">--save</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"attr\">pages:</span></span><br><span class=\"line\"> <span class=\"attr\">stage:</span> <span class=\"string\">deploy</span></span><br><span class=\"line\"> <span class=\"attr\">script:</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">hexo</span> <span class=\"string\">clean</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">hexo</span> <span class=\"string\">generate</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">hexo</span> <span class=\"string\">deploy</span></span><br><span class=\"line\"> <span class=\"attr\">artifacts:</span></span><br><span class=\"line\"> <span class=\"attr\">paths:</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">public</span></span><br><span class=\"line\"> <span class=\"attr\">only:</span></span><br><span class=\"line\"> <span class=\"bullet\">-</span> <span class=\"string\">master</span></span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n<h2 id=\"GitHub域名\"><a href=\"#GitHub域名\" class=\"headerlink\" title=\"GitHub域名\"></a>GitHub域名</h2><p>在 <code>source</code> 文件夹新建文件 <code>CNAME</code> 并写入你的域名，例如</p>\n<figure class=\"highlight plain\"><table><tr><td class=\"code\"><pre><span class=\"line\">www.holmeyoung.com</span><br></pre></td></tr></table></figure>\n<h2 id=\"总览\"><a href=\"#总览\" class=\"headerlink\" title=\"总览\"></a>总览</h2><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">$ tree -L 2 </span><br><span class=\"line\">.</span><br><span class=\"line\">├── _config.yml</span><br><span class=\"line\">├── package.json</span><br><span class=\"line\">├── scaffolds</span><br><span class=\"line\">│ ├── draft.md</span><br><span class=\"line\">│ ├── page.md</span><br><span class=\"line\">│ └── post.md</span><br><span class=\"line\">├── <span class=\"built_in\">source</span></span><br><span class=\"line\">│ ├── about</span><br><span class=\"line\">│ ├── categories</span><br><span class=\"line\">│ ├── CNAME</span><br><span class=\"line\">│ ├── _posts</span><br><span class=\"line\">│ └── tags</span><br><span class=\"line\">└── themes</span><br><span class=\"line\"> └── next</span><br></pre></td></tr></table></figure>\n<p>现在只需要写好 <code>markdown</code> 文件，放到 <code>_posts</code> 文件夹，提交到 <code>GitLab</code> ，稍等编译结束，访问你的博客域名就能看见啦</p>\n","categories":["安装记录"],"tags":["hexo"]},{"title":"centos7 配置 redis 主从集群 + 哨兵模式","url":"https://www.holmeyoung.com/blog/centos7-redis-cluster-sentinel/","content":"<h2 id=\"搭建环境\"><a href=\"#搭建环境\" class=\"headerlink\" title=\"搭建环境\"></a>搭建环境</h2><p>三台CentOS7服务器</p>\n<h2 id=\"实现目标\"><a href=\"#实现目标\" class=\"headerlink\" title=\"实现目标\"></a>实现目标</h2><p>在3台CentOS7服务器上搭建一主两从的redis集群，且配置哨兵模式，当master宕机，另外两个slave会选举出一个新的master，实现高可用。</p>\n<table>\n<thead>\n<tr>\n<th>master</th>\n<th>slave0</th>\n<th>slave1</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>ip0</td>\n<td>ip1</td>\n<td>ip2</td>\n</tr>\n</tbody>\n</table>\n<h2 id=\"准备工作\"><a href=\"#准备工作\" class=\"headerlink\" title=\"准备工作\"></a>准备工作</h2><h3 id=\"安装额外的软件包\"><a href=\"#安装额外的软件包\" class=\"headerlink\" title=\"安装额外的软件包\"></a>安装额外的软件包</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">$ yum install epel-release</span><br></pre></td></tr></table></figure>\n<h3 id=\"安装-redis\"><a href=\"#安装-redis\" class=\"headerlink\" title=\"安装 redis\"></a>安装 redis</h3><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> yum install redis</span></span><br></pre></td></tr></table></figure>\n<h2 id=\"配置一主两从\"><a href=\"#配置一主两从\" class=\"headerlink\" title=\"配置一主两从\"></a>配置一主两从</h2><h3 id=\"master（ip0）\"><a href=\"#master（ip0）\" class=\"headerlink\" title=\"master（ip0）\"></a>master（ip0）</h3><ul>\n<li><p>修改 <code>/etc/redis.conf</code></p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">$ vim /etc/redis.conf</span><br></pre></td></tr></table></figure>\n<p> 找到一下内容并修改</p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"comment\"># 将 127.0.0.1 修改为本机的IP（ip0）</span></span><br><span class=\"line\"><span class=\"built_in\">bind</span> ip0</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># 将 no 改为 yes，以守护进程的形式运行</span></span><br><span class=\"line\">daemonize yes</span><br></pre></td></tr></table></figure>\n<p> 修改完毕</p>\n</li>\n</ul>\n<ul>\n<li><p>测试</p>\n<p> 登录 <code>redis</code> </p>\n <figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip0</span></span><br></pre></td></tr></table></figure>\n<p> 列出所有 <code>kyes *</code></p>\n <figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip0</span></span><br><span class=\"line\">ip0:6379> keys *</span><br></pre></td></tr></table></figure>\n<p> 设置 <code>key</code> <code>value</code></p>\n <figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip0</span></span><br><span class=\"line\">ip0:6379> set name holmeyoung</span><br></pre></td></tr></table></figure>\n<p> 得到 <code>key</code> 的 <code>value</code></p>\n <figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip0</span></span><br><span class=\"line\">ip0:6379> get name</span><br></pre></td></tr></table></figure>\n<p> 删除 <code>key</code> </p>\n <figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip0</span></span><br><span class=\"line\">ip0:6379> del name</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n<h3 id=\"slave（ip1、ip2）\"><a href=\"#slave（ip1、ip2）\" class=\"headerlink\" title=\"slave（ip1、ip2）\"></a>slave（ip1、ip2）</h3><ul>\n<li><p>修改 <code>/etc/redis.conf</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> vim /etc/redis.conf</span></span><br></pre></td></tr></table></figure>\n<p>找到一下内容并修改</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"comment\"># 将 127.0.0.1 修改为本机的IP（eg：ip1）</span></span><br><span class=\"line\"><span class=\"built_in\">bind</span> ip1</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># 将 no 改为 yes，以守护进程的形式运行</span></span><br><span class=\"line\">daemonize yes</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># 增加该 slave节点 所要从的 master节点 的 ip/port</span></span><br><span class=\"line\"><span class=\"comment\"># 找到 # slaveof <masterip> <masterport></span></span><br><span class=\"line\"><span class=\"comment\"># 在下面增加一行</span></span><br><span class=\"line\">slaveof ip0 6379</span><br></pre></td></tr></table></figure>\n<p>修改完毕</p>\n</li>\n</ul>\n<ul>\n<li><p>测试</p>\n<p>同 <code>master</code> </p>\n</li>\n</ul>\n<h3 id=\"启动\"><a href=\"#启动\" class=\"headerlink\" title=\"启动\"></a>启动</h3><p>分别运行三台机器的 <code>redis-server</code> </p>\n<blockquote>\n<p>先运行 <code>master</code> </p>\n</blockquote>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-server /etc/redis.conf</span></span><br></pre></td></tr></table></figure>\n<h3 id=\"测试\"><a href=\"#测试\" class=\"headerlink\" title=\"测试\"></a>测试</h3><ul>\n<li><p>在 <code>master</code> 上</p>\n<p>连接 <code>redis</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip0</span></span><br></pre></td></tr></table></figure>\n<p>打印信息</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">ip0:6379> info Replication</span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> Replication</span></span><br><span class=\"line\">role:master</span><br><span class=\"line\">connected_slaves:2</span><br><span class=\"line\">slave0:ip=ip1,port=6379,state=online,offset=3242902,lag=1</span><br><span class=\"line\">slave1:ip=ip2,port=6379,state=online,offset=3243317,lag=0</span><br><span class=\"line\">master_repl_offset:3243317</span><br><span class=\"line\">repl_backlog_active:1</span><br><span class=\"line\">repl_backlog_size:1048576</span><br><span class=\"line\">repl_backlog_first_byte_offset:2194742</span><br><span class=\"line\">repl_backlog_histlen:1048576</span><br></pre></td></tr></table></figure>\n<p>写入一组 <code>key</code> <code>value</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">ip0:6379> set name holmeyoung</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n<ul>\n<li><p>在 <code>slave0</code> 上</p>\n<p>连接 <code>redis</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip1</span></span><br></pre></td></tr></table></figure>\n<p>读取数值</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">ip1:6379> get name</span><br></pre></td></tr></table></figure>\n<p>输出 <code>holmeyoung</code> 则正常</p>\n<blockquote>\n<p>因为 <code>slave</code> 只有读权限，所以是不可以写入数据的</p>\n</blockquote>\n</li>\n</ul>\n<ul>\n<li><p>在 <code>slave1</code> 上</p>\n<p>同 <code>slave0</code> </p>\n</li>\n</ul>\n<h2 id=\"配置哨兵模式\"><a href=\"#配置哨兵模式\" class=\"headerlink\" title=\"配置哨兵模式\"></a>配置哨兵模式</h2><h3 id=\"配置\"><a href=\"#配置\" class=\"headerlink\" title=\"配置\"></a>配置</h3><blockquote>\n<p>三台机器配置相同</p>\n</blockquote>\n<p>修改 <code>/etc/redis-sentinel.conf</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> vim /etc/redis-sentinel.conf</span></span><br></pre></td></tr></table></figure>\n<p>增改以下内容</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 保护模式，这个必须选择no，不然当master宕机，slave无法自动转换为master</span></span><br><span class=\"line\">protected-mode no</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 以守护进程运行</span></span><br><span class=\"line\">daemonize yes</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 日志文件位置</span></span><br><span class=\"line\">logfile \"/var/log/redis/sentinel.log\"</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 哨兵监控的master，主从配置一样，这里只用输入redis主节点的ip/port和法定人数</span></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 其中2是quorum，即权重，代表至少需要两个Sentinel节点认为主节点主观下线，才可判定主节点为客观下线。一般建议将其设置为Sentinel节点的一半加1。不仅如此，quorum还与Sentinel节点的领导者选举有关。为了选出Sentinel的领导者，至少需要max(quorum, num(sentinels) / 2 + 1)个Sentinel节点参与选举。</span></span><br><span class=\"line\">sentinel monitor mymaster ip0 6379 2</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> master或slave多长时间（默认30秒）不能使用后标记为s_down状态</span></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 每个Sentinel节点都要通过定期发送ping命令来判断Redis节点和其余Sentinel节点是否可达。</span></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 如果在指定的时间内，没有收到主节点的有效回复，则判断其为主观下线。需要注意的是，该参数不仅用来判断主节点状态，同样也用来判断该主节点下面的从节点及其它Sentinel的状态。其默认值为30s。</span></span><br><span class=\"line\">sentinel down-after-milliseconds mymaster 3000</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 在failover期间，允许多少个slave同时指向新的主节点。如果numslaves设置较大的话，虽然复制操作并不会阻塞主节点，但多个节点同时指向新的主节点，会增加主节点的网络和磁盘IO负载</span></span><br><span class=\"line\">sentinel parallel-syncs mymaster 1</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 定义故障切换超时时间。默认180000，单位秒，即3min。需要注意的是，该时间不是总的故障切换的时间，而是适用于故障切换的多个场景。</span></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 若sentinel在该配置值内未能完成failover操作（即故障时master/slave自动切换），则认为本次failover失败</span></span><br><span class=\"line\">sentinel failover-timeout mymaster 10000</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>注意 <code>sentinel monitor mymaster ip0 6379 2</code> 里面的 <code>ip0</code> 为 <code>master</code> 的IP</p>\n</blockquote>\n<h3 id=\"配置备份\"><a href=\"#配置备份\" class=\"headerlink\" title=\"配置备份\"></a>配置备份</h3><blockquote>\n<p>将配置文件 <code>/etc/redis.conf</code> 以及 <code>/etc/redis-sentinel.conf</code> 备份一份，因为当 <code>master</code> 宕机并唤醒哨兵模式后，系统会重写配置文件</p>\n</blockquote>\n<h3 id=\"启动-1\"><a href=\"#启动-1\" class=\"headerlink\" title=\"启动\"></a>启动</h3><p>分别运行三台机器的 <code>redis-sentinel</code> </p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-sentinel /etc/redis-sentinel.conf</span></span><br></pre></td></tr></table></figure>\n<h3 id=\"测试-1\"><a href=\"#测试-1\" class=\"headerlink\" title=\"测试\"></a>测试</h3><p>在 <code>master</code> 上杀死 <code>redis</code> 进程</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> pkill redis</span></span><br></pre></td></tr></table></figure>\n<p>随便登录一个 <code>slave</code> 服务器</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">$</span><span class=\"bash\"> redis-cli -h ip1</span></span><br></pre></td></tr></table></figure>\n<p>打印信息</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">ip0:6379> info Replication</span><br></pre></td></tr></table></figure>\n<p>会发现此时的 <code>master</code> 以及 <code>slave</code> 发生了更改</p>\n<h2 id=\"Reference\"><a href=\"#Reference\" class=\"headerlink\" title=\"Reference\"></a>Reference</h2><p><a href=\"https://www.cnblogs.com/ddrsql/p/6686282.html\" target=\"_blank\" rel=\"noopener\">CentOS7 配置 Redis Sentinel 主从集群配置</a></p>\n<p><a href=\"https://www.centos.bz/2017/07/centos7-redis-master-slave-replicate/\" target=\"_blank\" rel=\"noopener\">CentOS7 配置 Redis 主从复制方法</a></p>\n<h2 id=\"DONE\"><a href=\"#DONE\" class=\"headerlink\" title=\"DONE\"></a>DONE</h2>","categories":["安装记录"],"tags":["linux","redis"]},{"title":"centos7安装p40显卡驱动 + cuda10.0 + cudnn10.0","url":"https://www.holmeyoung.com/blog/centos7-p40-cuda10.0-cudnn10.0/","content":"<h2 id=\"环境\"><a href=\"#环境\" class=\"headerlink\" title=\"环境\"></a>环境</h2><ul>\n<li><p>系统</p>\n<p>CentOS Linux release 7.6.1810 (Core)</p>\n</li>\n<li><p>内核</p>\n<p>3.10.0-957.5.1.el7.x86_64</p>\n</li>\n<li><p>显卡</p>\n<p>NVIDIA Tesla P40</p>\n</li>\n<li><p>cuda</p>\n<p>cuda_10.0.130_410.48_linux.run</p>\n</li>\n<li><p>cudnn</p>\n<p>cudnn-10.0-linux-x64-v7.5.0.56.tgz</p>\n</li>\n</ul>\n<h2 id=\"显卡驱动\"><a href=\"#显卡驱动\" class=\"headerlink\" title=\"显卡驱动\"></a>显卡驱动</h2><h3 id=\"验证你的系统是否能识别上面的GPU\"><a href=\"#验证你的系统是否能识别上面的GPU\" class=\"headerlink\" title=\"验证你的系统是否能识别上面的GPU\"></a>验证你的系统是否能识别上面的GPU</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">lspci | grep -i nvidia</span><br></pre></td></tr></table></figure>\n<p>正常应该显示Nvidia显卡的型号，没有任何显示需要更新pci硬件库<code>update-pciids</code></p>\n<h3 id=\"验证系统是否是受支持的Linux版本\"><a href=\"#验证系统是否是受支持的Linux版本\" class=\"headerlink\" title=\"验证系统是否是受支持的Linux版本\"></a>验证系统是否是受支持的Linux版本</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">uname -m && cat /etc/*release</span><br></pre></td></tr></table></figure>\n<p>到这里查看受支持的Linux版本：<a href=\"http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements\" target=\"_blank\" rel=\"noopener\">http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements</a></p>\n<h3 id=\"验证系统是否有GCC编译环境\"><a href=\"#验证系统是否有GCC编译环境\" class=\"headerlink\" title=\"验证系统是否有GCC编译环境\"></a>验证系统是否有GCC编译环境</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo yum install gcc</span><br></pre></td></tr></table></figure>\n<h3 id=\"验证系统是否安装了正确的内核头文件和开发包\"><a href=\"#验证系统是否安装了正确的内核头文件和开发包\" class=\"headerlink\" title=\"验证系统是否安装了正确的内核头文件和开发包\"></a>验证系统是否安装了正确的内核头文件和开发包</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo yum install kernel-devel-$(uname -r) kernel-headers-$(uname -r)</span><br></pre></td></tr></table></figure>\n<h3 id=\"禁用nouveau\"><a href=\"#禁用nouveau\" class=\"headerlink\" title=\"禁用nouveau\"></a>禁用nouveau</h3><p>nouveau是一个第三方开源的Nvidia驱动，一般Linux安装的时候默认会安装这个驱动。 这个驱动会与Nvidia官方的驱动冲突，在安装Nvidia驱动和和CUDA之前应先禁用nouveau</p>\n<p>查看系统是否正在使用nouveau</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">lsmod | grep nouveau</span><br></pre></td></tr></table></figure>\n<p>如果有显示内容，则进行以下的步骤： <strong>Centos7禁用方法</strong></p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"comment\">#新建一个配置文件</span></span><br><span class=\"line\">sudo vim /etc/modprobe.d/blacklist.conf</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#写入以下内容</span></span><br><span class=\"line\">blacklist nouveau</span><br><span class=\"line\">options nouveau modeset=0</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#保存并退出</span></span><br><span class=\"line\">:wq</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#备份当前的镜像</span></span><br><span class=\"line\">sudo mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#建立新的镜像</span></span><br><span class=\"line\">sudo dracut /boot/initramfs-$(uname -r).img $(uname -r)</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#重启</span></span><br><span class=\"line\">sudo reboot</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#最后输入上面的命令验证</span></span><br><span class=\"line\">lsmod | grep nouveau</span><br></pre></td></tr></table></figure>\n<p>总之最后一定要保证<code>lsmod | grep nouveau</code>命令没有任何输出内容。 这也是runfile文件安装方式相对rpm/deb方式麻烦的一点，如果本节禁用nouveau有问题，可以尝试rpm/deb方式安装，rpm/deb方式安装会自动禁用</p>\n<h3 id=\"下载驱动\"><a href=\"#下载驱动\" class=\"headerlink\" title=\"下载驱动\"></a>下载驱动</h3><p>首先在官网 <a href=\"https://www.nvidia.com/Download/index.aspx?lang=cn\" target=\"_blank\" rel=\"noopener\">https://www.nvidia.com/Download/index.aspx?lang=cn</a> 查找自己显卡的驱动</p>\n<p>我这里是 <code>NVIDIA Tesla P40</code> ，因为列表中可支持的最新版本cuda是 <code>10.0</code> ，所以下载的cuda为 10.0 的版本：<strong>NVIDIA-Linux-x86_64-410.104.run</strong></p>\n<h3 id=\"安装驱动\"><a href=\"#安装驱动\" class=\"headerlink\" title=\"安装驱动\"></a>安装驱动</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo sh NVIDIA-Linux-x86_64-410.104.run --kernel-source-path=/usr/src/kernels/3.10.0-957.5.1.el7.x86_64</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>这里必须指明 <code>kernel-source-path</code> ，否则安装过程会报错</p>\n</blockquote>\n<h3 id=\"检验安装\"><a href=\"#检验安装\" class=\"headerlink\" title=\"检验安装\"></a>检验安装</h3><p>执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">nvidia-smi</span><br></pre></td></tr></table></figure>\n<p>如下输出则说明成功</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">Thu Mar 14 18:40:38 2019 </span><br><span class=\"line\">+-----------------------------------------------------------------------------+</span><br><span class=\"line\">| NVIDIA-SMI 410.104 Driver Version: 410.104 CUDA Version: 10.0 |</span><br><span class=\"line\">|-------------------------------+----------------------+----------------------+</span><br><span class=\"line\">| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |</span><br><span class=\"line\">| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |</span><br><span class=\"line\">|===============================+======================+======================|</span><br><span class=\"line\">| 0 Tesla P40 Off | 00000000:3B:00.0 Off | 0 |</span><br><span class=\"line\">| N/A 25C P0 47W / 250W | 0MiB / 22919MiB | 0% Default |</span><br><span class=\"line\">+-------------------------------+----------------------+----------------------+</span><br><span class=\"line\"> </span><br><span class=\"line\">+-----------------------------------------------------------------------------+</span><br><span class=\"line\">| Processes: GPU Memory |</span><br><span class=\"line\">| GPU PID Type Process name Usage |</span><br><span class=\"line\">|=============================================================================|</span><br><span class=\"line\">| No running processes found |</span><br><span class=\"line\">+-----------------------------------------------------------------------------+</span><br></pre></td></tr></table></figure>\n<h2 id=\"cuda\"><a href=\"#cuda\" class=\"headerlink\" title=\"cuda\"></a>cuda</h2><h3 id=\"下载cuda\"><a href=\"#下载cuda\" class=\"headerlink\" title=\"下载cuda\"></a>下载cuda</h3><p>因为 NVIDIA 官网上，P40的cuda最多支持到10.0版本，驱动下载的也是10.0版本，所以这里下载<code>cuda10.0</code></p>\n<p>各版本下载链接 <a href=\"https://developer.nvidia.com/cuda-toolkit-archive\" target=\"_blank\" rel=\"noopener\">https://developer.nvidia.com/cuda-toolkit-archive</a></p>\n<p>我这里下载的是 <code>runfile</code></p>\n<h3 id=\"安装cuda\"><a href=\"#安装cuda\" class=\"headerlink\" title=\"安装cuda\"></a>安装cuda</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo sh cuda_10.0.130_410.48_linux.run --kernel-source-path=/usr/src/kernels/3.10.0-957.5.1.el7.x86_64</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>这里必须指明 <code>kernel-source-path</code> ，否则安装过程会报错</p>\n</blockquote>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"comment\">#刚开始安装会进入more模式，一直按`空格`即可</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#Do you accept the previously read EULA?</span></span><br><span class=\"line\"><span class=\"comment\">#是否接受协议</span></span><br><span class=\"line\">accept</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 387.26?</span></span><br><span class=\"line\"><span class=\"comment\">#是否安装Nvida驱动，不需要</span></span><br><span class=\"line\">n</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#Install the CUDA 10.0 Toolkit?</span></span><br><span class=\"line\">y</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#Enter Toolkit Location</span></span><br><span class=\"line\"><span class=\"comment\">#输入Toolkit的安装目录</span></span><br><span class=\"line\"><span class=\"comment\">#一般默认即可</span></span><br><span class=\"line\"><span class=\"comment\">#如果更改的话，注意是不可以改成 /usr/local/cuda 的，其余都可以</span></span><br><span class=\"line\">回车</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#Do you want to install a symbolic link at /usr/local/cuda?</span></span><br><span class=\"line\"><span class=\"comment\">#创建一个软连接，我选择是</span></span><br><span class=\"line\">y</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#Install the CUDA 10.0 Samples?</span></span><br><span class=\"line\"><span class=\"comment\">#安装CUDA官方示例包</span></span><br><span class=\"line\"><span class=\"comment\">#这里选择 no，因为安装目录下就有</span></span><br><span class=\"line\">n</span><br><span class=\"line\"></span><br><span class=\"line\"></span><br><span class=\"line\"></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\">#如果版本不同，提示也可能不同，根据情况输入或选择</span></span><br></pre></td></tr></table></figure>\n<h2 id=\"cudnn\"><a href=\"#cudnn\" class=\"headerlink\" title=\"cudnn\"></a>cudnn</h2><h3 id=\"下载cudnn\"><a href=\"#下载cudnn\" class=\"headerlink\" title=\"下载cudnn\"></a>下载cudnn</h3><blockquote>\n<p>注意和 <code>cuda</code> 版本匹配</p>\n</blockquote>\n<h3 id=\"安装cudnn\"><a href=\"#安装cudnn\" class=\"headerlink\" title=\"安装cudnn\"></a>安装cudnn</h3><p>按照官网命令</p>\n<p><a href=\"https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html\" target=\"_blank\" rel=\"noopener\">https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html</a></p>\n<ul>\n<li><p>Navigate to your <code>cudnnpath</code> directory containing the cuDNN Tar file.</p>\n</li>\n<li><p>Unzip the cuDNN package.</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">tar -xzvf cudnn-10.0-linux-x64-v7.5.0.56.tgz</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>Copy the following files into the CUDA Toolkit directory, and change the file permissions.</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo cp cuda/include/cudnn.h /usr/<span class=\"built_in\">local</span>/cuda/include</span><br><span class=\"line\">sudo cp cuda/lib64/libcudnn* /usr/<span class=\"built_in\">local</span>/cuda/lib64</span><br><span class=\"line\">sudo chmod a+r /usr/<span class=\"built_in\">local</span>/cuda/include/cudnn.h /usr/<span class=\"built_in\">local</span>/cuda/lib64/libcudnn*</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n","categories":["安装记录"],"tags":["linux","gpu","机器学习"]},{"title":"centos7 anaconda3 安装 youcompleteme","url":"https://www.holmeyoung.com/blog/centos7-anaconda3-youcompleteme/","content":"<h3 id=\"环境\"><a href=\"#环境\" class=\"headerlink\" title=\"环境\"></a>环境</h3><ul>\n<li><strong><code>CentOS7</code></strong></li>\n<li><strong><code>Anaconda3</code></strong></li>\n<li><strong><code>oh-my-zsh</code></strong></li>\n</ul>\n<h3 id=\"更新yum\"><a href=\"#更新yum\" class=\"headerlink\" title=\"更新yum\"></a>更新yum</h3><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo yum upgrade</span><br><span class=\"line\">sudo yum update</span><br></pre></td></tr></table></figure>\n<h3 id=\"安装可能需要的\"><a href=\"#安装可能需要的\" class=\"headerlink\" title=\"安装可能需要的\"></a>安装可能需要的</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo yum install python-devel</span><br><span class=\"line\">sudo yum install cmake</span><br></pre></td></tr></table></figure>\n<h3 id=\"升级gcc\"><a href=\"#升级gcc\" class=\"headerlink\" title=\"升级gcc\"></a>升级gcc</h3><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo yum install centos-release-scl -y</span><br><span class=\"line\">sudo yum install devtoolset-3-toolchain -y</span><br><span class=\"line\">sudo yum install gcc-c++</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 如果安装了 oh-my—zsh 则不执行下面这句，否则会切换为bash</span></span><br><span class=\"line\">sudo scl enable devtoolset-3 bash</span><br></pre></td></tr></table></figure>\n<h3 id=\"禁用Anaconda3\"><a href=\"#禁用Anaconda3\" class=\"headerlink\" title=\"禁用Anaconda3\"></a>禁用Anaconda3</h3><p>将 </p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">export</span> PATH=<span class=\"string\">\"/root/anaconda3/bin:<span class=\"variable\">$PATH</span>\"</span></span><br></pre></td></tr></table></figure>\n<p>从 <code>.zshrc</code> 、<code>.bashrc</code> 中注释掉</p>\n<p>注释后<strong>根目录下</strong>执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">source</span> .zshrc</span><br><span class=\"line\"><span class=\"built_in\">source</span> .bashrc</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>此时执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> python -V</span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>依旧会显示 <code>Python 3.6.5 :: Anaconda, Inc.</code> </p>\n<p><strong>所以我们需要新开一个终端运行下面的命令</strong></p>\n</blockquote>\n<h3 id=\"卸载系统原生vim（vim7）\"><a href=\"#卸载系统原生vim（vim7）\" class=\"headerlink\" title=\"卸载系统原生vim（vim7）\"></a>卸载系统原生vim（vim7）</h3><figure class=\"highlight plain\"><table><tr><td class=\"code\"><pre><span class=\"line\">yum remove vim</span><br></pre></td></tr></table></figure>\n<h3 id=\"自己源码安装vim（vim8）\"><a href=\"#自己源码安装vim（vim8）\" class=\"headerlink\" title=\"自己源码安装vim（vim8）\"></a>自己源码安装vim（vim8）</h3><h4 id=\"安装命令\"><a href=\"#安装命令\" class=\"headerlink\" title=\"安装命令\"></a>安装命令</h4><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">git clone https://github.com/vim/vim.git</span><br><span class=\"line\">cd vim/src</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p><strong><code>下面是最关键的一步，多少次失败都是因为这个命令不对！！！</code></strong></p>\n</blockquote>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">./configure --with-features=huge --<span class=\"built_in\">enable</span>-pythoninterp=yes --<span class=\"built_in\">enable</span>-python3interp=yes --with-python3-config-dir=/root/anaconda3/lib/python3.6/config-3.6m-x86_64-linux-gnu</span><br></pre></td></tr></table></figure>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo make</span><br><span class=\"line\">sudo make install</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p><code>--with-python3-config-dir=/root/anaconda3/lib/python3.6/config-3.6m-x86_64-linux-gnu</code></p>\n<p>这条命令根据不同人 <code>anaconda3</code> 安装位置不同会有变动</p>\n</blockquote>\n<h4 id=\"环境设置\"><a href=\"#环境设置\" class=\"headerlink\" title=\"环境设置\"></a>环境设置</h4><p>将</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">export</span> PATH=<span class=\"string\">\"/usr/local/bin:<span class=\"variable\">$PATH</span>\"</span></span><br></pre></td></tr></table></figure>\n<p>配置了 <code>oh-my-zsh</code> 的，添加到 <code>.zshrc</code></p>\n<p>没配置 <code>oh-my-zsh</code> 的，添加到 <code>.bashrc</code></p>\n<p>添加后<strong>根目录下</strong>执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">source</span> .zshrc</span><br><span class=\"line\"><span class=\"comment\"># source .bashrc</span></span><br></pre></td></tr></table></figure>\n<h4 id=\"关于安装\"><a href=\"#关于安装\" class=\"headerlink\" title=\"关于安装\"></a>关于安装</h4><h5 id=\"命令参数\"><a href=\"#命令参数\" class=\"headerlink\" title=\"命令参数\"></a>命令参数</h5><p>这里列举一些<code>./configure</code> 后面的配置选项</p>\n<table>\n<thead>\n<tr>\n<th>参数</th>\n<th>含义</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td><strong><code>--with-features=huge</code></strong></td>\n<td>支持最大特性</td>\n</tr>\n<tr>\n<td><strong><code>--enable-rubyinterp=yes</code></strong></td>\n<td>打开对ruby编写的插件的支持</td>\n</tr>\n<tr>\n<td><strong><code>--enable-pythoninterp=yes</code></strong></td>\n<td>打开对python编写的插件的支持</td>\n</tr>\n<tr>\n<td><strong><code>--enable-python3interp=yes</code></strong></td>\n<td>打开对python3编写的插件的支持</td>\n</tr>\n<tr>\n<td><strong><code>--enable-luainterp=yes</code></strong></td>\n<td>打开对lua编写的插件的支持</td>\n</tr>\n<tr>\n<td><strong><code>--enable-perlinterp=yes</code></strong></td>\n<td>打开对perl编写的插件的支持</td>\n</tr>\n<tr>\n<td><strong><code>--enable-multibyte=yes</code></strong></td>\n<td>打开多字节支持，可以在Vim中输入中文</td>\n</tr>\n<tr>\n<td><strong><code>--enable-cscope=yes</code></strong></td>\n<td>打开对cscope的支持</td>\n</tr>\n<tr>\n<td><strong><code>--with-python-config-dir=/usr/lib/python2.7/config-x86_64-linux-gnu</code></strong></td>\n<td>指定python 路径</td>\n</tr>\n<tr>\n<td><strong><code>--with-python3-config-dir=/usr/lib/python3.5/config-3.5m-x86_64-linux-gnu</code></strong></td>\n<td>指定python3路径</td>\n</tr>\n<tr>\n<td><strong><code>--prefix=/usr/local/vim</code></strong></td>\n<td>指定将要安装到的路径(自行创建)</td>\n</tr>\n</tbody>\n</table>\n<h5 id=\"安装路径\"><a href=\"#安装路径\" class=\"headerlink\" title=\"安装路径\"></a>安装路径</h5><ul>\n<li><p>通过 <code>yum install xxx</code> 安装的 <strong>软件</strong> 在 <code>usr/bin/</code> 下</p>\n</li>\n<li><p>通过 <strong>源码</strong> 安装的 <strong>软件</strong> 在 <code>usr/local/bin/</code> 下</p>\n</li>\n</ul>\n<p>当我们通过 <code>yum remove vim</code> 卸载 <code>vim</code> 后再通过源码安装，安装好之后再输入 <code>vim</code> 默认指向的就是 <code>usr/local/bin/</code> 下的自己安装的</p>\n<p>如下</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">➜ /root whereis vim</span><br><span class=\"line\">vim: /usr/local/bin/vim /usr/share/vim /usr/share/man/man1/vim.1.gz</span><br></pre></td></tr></table></figure>\n<h3 id=\"安装YouCompleteMe\"><a href=\"#安装YouCompleteMe\" class=\"headerlink\" title=\"安装YouCompleteMe\"></a>安装YouCompleteMe</h3><p>根目录下执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git <span class=\"built_in\">clone</span> https://github.com/VundleVim/Vundle.vim.git ~/.vim/bundle/Vundle.vim</span><br></pre></td></tr></table></figure>\n<p>编辑 <code>.vimrc</code> ，加入</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"string\">\" Vundle Setting</span></span><br><span class=\"line\"><span class=\"string\">set nocompatible</span></span><br><span class=\"line\"><span class=\"string\">filetype off</span></span><br><span class=\"line\"><span class=\"string\">set rtp+=~/.vim/bundle/Vundle.vim</span></span><br><span class=\"line\"><span class=\"string\">call vundle#begin()</span></span><br><span class=\"line\"><span class=\"string\"></span></span><br><span class=\"line\"><span class=\"string\">\"</span> <span class=\"built_in\">let</span> Vundle manage Vundle, required</span><br><span class=\"line\">Plugin <span class=\"string\">'VundleVim/Vundle.vim'</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"string\">\" Plugins</span></span><br><span class=\"line\"><span class=\"string\">Plugin 'Valloric/YouCompleteMe'</span></span><br><span class=\"line\"><span class=\"string\"></span></span><br><span class=\"line\"><span class=\"string\">\"</span> All of your Plugins must be added before the following line</span><br><span class=\"line\">call vundle<span class=\"comment\">#end()</span></span><br><span class=\"line\">filetype plugin indent on</span><br></pre></td></tr></table></figure>\n<p>保存退出</p>\n<p>在终端执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">vim +PluginInstall +qall</span><br></pre></td></tr></table></figure>\n<p>或者在 <code>vim</code> 中执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">:PluginInstall</span><br></pre></td></tr></table></figure>\n<p>等待插件安装完成自动退出</p>\n<p>此时会提示 <code>YouCompleteMe</code> 未编译</p>\n<p>执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">cd</span> ~/.vim/bundle/YouCompleteMe</span><br><span class=\"line\">./install.py --clang-completer</span><br></pre></td></tr></table></figure>\n<p>这里只编译了 <code>YouCompleteMe</code> 对 <code>C</code> 族的补全，更多语言的补全命令可见 <a href=\"https://github.com/Valloric/YouCompleteMe\" target=\"_blank\" rel=\"noopener\">YouCompleteMe</a> </p>\n<h4 id=\"关于另一个版本的安装方法\"><a href=\"#关于另一个版本的安装方法\" class=\"headerlink\" title=\"关于另一个版本的安装方法\"></a>关于另一个版本的安装方法</h4><ol>\n<li><p>如果已经安装 <code>Vundle</code> 直接跳到 <code>2</code></p>\n<p>根目录下执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git <span class=\"built_in\">clone</span> https://github.com/VundleVim/Vundle.vim.git ~/.vim/bundle/Vundle.vim</span><br></pre></td></tr></table></figure>\n<p>编辑 <code>.vimrc</code> ，加入</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"string\">\" Vundle Setting</span></span><br><span class=\"line\"><span class=\"string\">set nocompatible</span></span><br><span class=\"line\"><span class=\"string\">filetype off</span></span><br><span class=\"line\"><span class=\"string\">set rtp+=~/.vim/bundle/Vundle.vim</span></span><br><span class=\"line\"><span class=\"string\">call vundle#begin()</span></span><br><span class=\"line\"><span class=\"string\"></span></span><br><span class=\"line\"><span class=\"string\">\"</span> <span class=\"built_in\">let</span> Vundle manage Vundle, required</span><br><span class=\"line\">Plugin <span class=\"string\">'VundleVim/Vundle.vim'</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"string\">\" Plugins</span></span><br><span class=\"line\"><span class=\"string\">\"</span> Plugin <span class=\"string\">'Valloric/YouCompleteMe'</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"string\">\" All of your Plugins must be added before the following line</span></span><br><span class=\"line\"><span class=\"string\">call vundle#end()</span></span><br><span class=\"line\"><span class=\"string\">filetype plugin indent on</span></span><br></pre></td></tr></table></figure>\n<p>保存退出执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">vim +PluginInstall +qall</span><br></pre></td></tr></table></figure>\n<p>或者在 <code>vim</code> 中执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">:PluginInstall</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>安装 <code>YouCompleteMe</code></p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git <span class=\"built_in\">clone</span> https://github.com/Valloric/YouCompleteMe.git ~/.vim/bundle/YouCompleteMe</span><br><span class=\"line\"><span class=\"built_in\">cd</span> ~/.vim/bundle/YouCompleteMe</span><br><span class=\"line\">git submodule update --init --recursive</span><br></pre></td></tr></table></figure>\n<p>编辑 <code>.vimrc</code> ，取消第 <code>11</code> 行的注释，即变为</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">Plugin <span class=\"string\">'Valloric/YouCompleteMe'</span></span><br></pre></td></tr></table></figure>\n<p>保存退出后执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">cd</span> ~/.vim/bundle/YouCompleteMe</span><br><span class=\"line\">./install.py --clang-completer</span><br></pre></td></tr></table></figure>\n<p><strong>DONE</strong></p>\n</li>\n</ol>\n<blockquote>\n<p><strong><code>这个安装方法我试了是没有成功的，可能是我环境的原因</code></strong></p>\n</blockquote>\n<blockquote>\n<p>亲测在 <code>.vimrc</code> 中添加</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> Plugin <span class=\"string\">'Valloric/YouCompleteMe'</span></span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>并执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> vim +PluginInstall +qall</span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>后</p>\n<p>命令</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> git submodule update --init --recursive</span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>会失效，故采用 <code>Vundle</code> 安装可直接执行命令</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> ./install.py --clang-completer</span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>进行编译</p>\n</blockquote>\n<h3 id=\"放生Anaconda3\"><a href=\"#放生Anaconda3\" class=\"headerlink\" title=\"放生Anaconda3\"></a>放生Anaconda3</h3><p>将</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">export</span> PATH=<span class=\"string\">\"/root/anaconda3/bin:<span class=\"variable\">$PATH</span>\"</span></span><br></pre></td></tr></table></figure>\n<p>在 <code>.zshrc</code> 中的注释去掉</p>\n<p>没有安装 <code>oh-my-zsh</code> 的则去掉 <code>.bashrc</code> 中的注释</p>\n<p>保存退出后<strong>根目录下</strong>执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">source</span> .zshrc</span><br><span class=\"line\"><span class=\"built_in\">source</span> .bashrc</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>此时执行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> python -V</span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>会显示 <code>Python 3.6.5 :: Anaconda, Inc.</code> </p>\n</blockquote>\n<p><strong>DONE</strong></p>\n<h3 id=\"Ref\"><a href=\"#Ref\" class=\"headerlink\" title=\"Ref\"></a>Ref</h3><blockquote>\n<p><a href=\"https://www.cnblogs.com/hugb/p/8593012.html\" target=\"_blank\" rel=\"noopener\">Centos7安装vim8.0 + YouCompleteMe</a></p>\n</blockquote>\n","categories":["安装记录"],"tags":["linux"]},{"title":"python 进程间管道通信","url":"https://www.holmeyoung.com/blog/python-process-fifo/","content":"<p>实现了两个 <code>Linux</code> 进程之间通过管道进行通讯的代码逻辑。分为 <code>client.py</code> 与 <code>server.py</code> 运行时需要先运行 <code>server</code> 再运行 <code>client</code></p>\n<p><code>client.py</code></p>\n<figure class=\"highlight python\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"keyword\">import</span> os</span><br><span class=\"line\"><span class=\"keyword\">import</span> time</span><br><span class=\"line\"></span><br><span class=\"line\">write_path = <span class=\"string\">\"pipe.in\"</span></span><br><span class=\"line\">read_path = <span class=\"string\">\"pipe.out\"</span></span><br><span class=\"line\"></span><br><span class=\"line\">wf = os.open(write_path, os.O_SYNC | os.O_CREAT | os.O_RDWR)</span><br><span class=\"line\">rf = <span class=\"literal\">None</span></span><br><span class=\"line\"><span class=\"comment\"># rf = os.open(read_path, os.O_RDONLY)</span></span><br><span class=\"line\"><span class=\"keyword\">for</span> i <span class=\"keyword\">in</span> range(<span class=\"number\">1</span>, <span class=\"number\">11</span>):</span><br><span class=\"line\"> msg = <span class=\"string\">\"msg \"</span> + str(i)</span><br><span class=\"line\"> msg = msg.encode()</span><br><span class=\"line\"> len_send = os.write(wf, msg)</span><br><span class=\"line\"> <span class=\"keyword\">print</span> (<span class=\"string\">\"sent msg: %s\"</span> % msg)</span><br><span class=\"line\"></span><br><span class=\"line\"> <span class=\"keyword\">if</span> rf <span class=\"keyword\">is</span> <span class=\"literal\">None</span>:</span><br><span class=\"line\"> rf = os.open(read_path, os.O_RDONLY)</span><br><span class=\"line\"></span><br><span class=\"line\"> s = os.read(rf, <span class=\"number\">1024</span>)</span><br><span class=\"line\"> <span class=\"keyword\">if</span> len(s) == <span class=\"number\">0</span>:</span><br><span class=\"line\"> <span class=\"keyword\">break</span></span><br><span class=\"line\"> <span class=\"keyword\">print</span> (<span class=\"string\">\"received msg: %s\"</span> % s)</span><br><span class=\"line\"></span><br><span class=\"line\"> time.sleep(<span class=\"number\">1</span>)</span><br><span class=\"line\"></span><br><span class=\"line\">os.write(wf, <span class=\"string\">'exit'</span>.encode())</span><br><span class=\"line\"></span><br><span class=\"line\">os.close(rf)</span><br><span class=\"line\">os.close(wf)</span><br></pre></td></tr></table></figure>\n<p><code>server.py</code></p>\n<figure class=\"highlight python\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"keyword\">import</span> os, time</span><br><span class=\"line\"></span><br><span class=\"line\">read_path = <span class=\"string\">\"pipe.in\"</span></span><br><span class=\"line\">write_path = <span class=\"string\">\"pipe.out\"</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"keyword\">if</span> os.path.exists(read_path):</span><br><span class=\"line\"> os.remove(read_path)</span><br><span class=\"line\"><span class=\"keyword\">if</span> os.path.exists(write_path):</span><br><span class=\"line\"> os.remove(write_path)</span><br><span class=\"line\"></span><br><span class=\"line\">os.mkfifo(write_path)</span><br><span class=\"line\">os.mkfifo(read_path)</span><br><span class=\"line\"></span><br><span class=\"line\">rf = os.open(read_path, os.O_RDONLY)</span><br><span class=\"line\">wf = os.open(write_path, os.O_SYNC | os.O_CREAT | os.O_RDWR)</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"keyword\">while</span> <span class=\"literal\">True</span>:</span><br><span class=\"line\"> s = os.read(rf, <span class=\"number\">1024</span>)</span><br><span class=\"line\"> <span class=\"keyword\">print</span> (<span class=\"string\">\"received msg: %s\"</span> % s)</span><br><span class=\"line\"> <span class=\"keyword\">if</span> len(s) == <span class=\"number\">0</span>:</span><br><span class=\"line\"> time.sleep(<span class=\"number\">1</span>)</span><br><span class=\"line\"> <span class=\"keyword\">continue</span></span><br><span class=\"line\"></span><br><span class=\"line\"> <span class=\"keyword\">if</span> <span class=\"string\">\"exit\"</span>.encode() <span class=\"keyword\">in</span> s:</span><br><span class=\"line\"> <span class=\"keyword\">break</span></span><br><span class=\"line\"></span><br><span class=\"line\"> os.write(wf, s)</span><br><span class=\"line\"></span><br><span class=\"line\">os.close(rf)</span><br><span class=\"line\">os.close(wf)</span><br></pre></td></tr></table></figure>","categories":["python"],"tags":["python"]},{"title":"修改 pip 源为清华大学开源镜像","url":"https://www.holmeyoung.com/blog/change-pip-mirror/","content":"<h3 id=\"pypi-镜像使用帮助\"><a href=\"#pypi-镜像使用帮助\" class=\"headerlink\" title=\"pypi 镜像使用帮助\"></a>pypi 镜像使用帮助</h3><p><code>pypi</code> 镜像每 5 分钟同步一次。</p>\n<h3 id=\"临时使用\"><a href=\"#临时使用\" class=\"headerlink\" title=\"临时使用\"></a>临时使用</h3><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">pip install -i https://pypi.tuna.tsinghua.edu.cn/simple some-package</span><br></pre></td></tr></table></figure>\n<p>注意， <code>simple</code> 不能少, 是 <code>https</code> 而不是 <code>http</code> </p>\n<h3 id=\"设为默认\"><a href=\"#设为默认\" class=\"headerlink\" title=\"设为默认\"></a>设为默认</h3><p>修改 <code>~/.config/pip/pip.conf</code> <strong>(Linux)</strong>, <code>%APPDATA%\\pip\\pip.ini</code> <strong>(Windows 10)</strong> 或 <code>$HOME/Library/Application Support/pip/pip.conf</code> <strong>(macOS)</strong> (没有就创建一个)， 修改 <code>index-url</code> 至 <code>tuna</code> ，例如</p>\n<figure class=\"highlight plain\"><table><tr><td class=\"code\"><pre><span class=\"line\">[global]</span><br><span class=\"line\">index-url = https://pypi.tuna.tsinghua.edu.cn/simple</span><br></pre></td></tr></table></figure>\n<p><code>pip</code> 和 <code>pip3</code> 并存时，只需修改 <code>~/.pip/pip.conf</code> </p>\n","categories":["linux"],"tags":["python"]},{"title":"ubuntu14.04 安装 caffe","url":"https://www.holmeyoung.com/blog/ubuntu14-caffe/","content":"<h3 id=\"组件-amp-版本\"><a href=\"#组件-amp-版本\" class=\"headerlink\" title=\"组件&版本\"></a>组件&版本</h3><ul>\n<li><p><strong>Ubuntu</strong></p>\n<ul>\n<li>14.04</li>\n</ul>\n</li>\n<li><p><strong>CUDA</strong></p>\n<ul>\n<li>8.0</li>\n</ul>\n</li>\n<li><strong>cuDNN</strong><ul>\n<li>cudnn-7.0-linux-x64-<strong>v3.0</strong>-prod.tgz</li>\n</ul>\n</li>\n<li><strong>opencv</strong><ul>\n<li>3.2</li>\n</ul>\n</li>\n<li><strong>python</strong><ul>\n<li>Python 2.7.6</li>\n</ul>\n</li>\n<li><strong>caffe</strong><ul>\n<li>release candidate 3</li>\n</ul>\n</li>\n</ul>\n<h3 id=\"安装过程\"><a href=\"#安装过程\" class=\"headerlink\" title=\"安装过程\"></a>安装过程</h3><h4 id=\"安装开发所需依赖包\"><a href=\"#安装开发所需依赖包\" class=\"headerlink\" title=\"安装开发所需依赖包\"></a>安装开发所需依赖包</h4><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo apt-get install build-essential </span><br><span class=\"line\">sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev libgflags-dev libgoogle-glog-dev liblmdb-dev protobuf-compiler</span><br></pre></td></tr></table></figure>\n<h4 id=\"安装CUDA-8-0\"><a href=\"#安装CUDA-8-0\" class=\"headerlink\" title=\"安装CUDA 8.0\"></a>安装CUDA 8.0</h4><ul>\n<li><p>确定 <code>GPU</code> 支持 <code>CUDA</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">lspci | grep -i nvidia</span><br></pre></td></tr></table></figure>\n<p>显示</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">02:00.0 VGA compatible controller: NVIDIA Corporation Device 1b06 (rev a1)</span><br><span class=\"line\">02:00.1 Audio device: NVIDIA Corporation Device 10ef (rev a1)</span><br><span class=\"line\">03:00.0 VGA compatible controller: NVIDIA Corporation Device 1b06 (rev a1)</span><br><span class=\"line\">03:00.1 Audio device: NVIDIA Corporation Device 10ef (rev a1)</span><br><span class=\"line\">82:00.0 VGA compatible controller: NVIDIA Corporation Device 1b06 (rev a1)</span><br><span class=\"line\">82:00.1 Audio device: NVIDIA Corporation Device 10ef (rev a1)</span><br><span class=\"line\">83:00.0 VGA compatible controller: NVIDIA Corporation Device 1b06 (rev a1)</span><br><span class=\"line\">83:00.1 Audio device: NVIDIA Corporation Device 10ef (rev a1)</span><br></pre></td></tr></table></figure>\n<p>根据 <strong><a href=\"http://developer.nvidia.com/cuda-gpus\" target=\"_blank\" rel=\"noopener\">http://developer.nvidia.com/cuda-gpus</a></strong> 去验证，发现支持 <code>CUDA</code></p>\n</li>\n<li><p>确定 <code>linux</code> 版本支持 <code>CUDA</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">uname -m && cat /etc/*release</span><br></pre></td></tr></table></figure>\n<p>显示</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">x86_64</span><br><span class=\"line\">DISTRIB_ID=Ubuntu</span><br><span class=\"line\">DISTRIB_RELEASE=14.04</span><br><span class=\"line\">DISTRIB_CODENAME=trusty</span><br><span class=\"line\">DISTRIB_DESCRIPTION=\"Ubuntu 14.04.5 LTS\"</span><br><span class=\"line\">NAME=\"Ubuntu\"</span><br><span class=\"line\">VERSION=\"14.04.5 LTS, Trusty Tahr\"</span><br><span class=\"line\">ID=ubuntu</span><br><span class=\"line\">ID_LIKE=debian</span><br><span class=\"line\">PRETTY_NAME=\"Ubuntu 14.04.5 LTS\"</span><br><span class=\"line\">VERSION_ID=\"14.04\"</span><br><span class=\"line\">HOME_URL=\"http://www.ubuntu.com/\"</span><br><span class=\"line\">SUPPORT_URL=\"http://help.ubuntu.com/\"</span><br><span class=\"line\">BUG_REPORT_URL=\"http://bugs.launchpad.net/ubuntu/\"</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>确定系统已经安装了 <code>gcc</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">gcc --version</span><br></pre></td></tr></table></figure>\n<p>显示</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">gcc (Ubuntu 4.8.5-2ubuntu1~14.04.1) 4.8.5</span><br><span class=\"line\">Copyright (C) 2015 Free Software Foundation, Inc.</span><br><span class=\"line\">This is free software; see the source for copying conditions. There is NO</span><br><span class=\"line\">warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>确定系统已经安装了正确的 <code>Kernel Headers</code> 和开发包</p>\n<ul>\n<li><p>查看系统正在运行的 <code>kernel</code> 版本</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">uname -r</span><br></pre></td></tr></table></figure>\n<p>显示</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">3.13.0-142-generic</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>安装对应的 <code>kernels header</code> 和开发包</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo apt-get install linux-headers-$(uname -r)</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n</li>\n<li><p>安装 <code>CUDA</code></p>\n<ul>\n<li>下载对应版本的cuda，从 <strong><a href=\"https://developer.nvidia.com/cuda-downloads\" target=\"_blank\" rel=\"noopener\">https://developer.nvidia.com/cuda-downloads</a></strong> 下载 <code>deb(network)</code> 版</li>\n</ul>\n</li>\n</ul>\n<ul>\n<li><p>进行 <code>md5</code> 校验</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">md5sum cuda-repo-ubuntu1404_8.0.61-1_amd64.deb</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>对比相同，然后使用 <code>deb</code> 文件进行安装</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo dpkg -i cuda-repo-ubuntu1404_8.0.61-1_amd64.deb</span><br><span class=\"line\">sudo apt-get update</span><br><span class=\"line\">sudo apt-get install cuda</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>重启，完成 <code>cuda</code> 安装</p>\n</li>\n</ul>\n<h4 id=\"安装-cuDNN\"><a href=\"#安装-cuDNN\" class=\"headerlink\" title=\"安装 cuDNN\"></a>安装 cuDNN</h4><ul>\n<li><p>从 <strong><a href=\"https://developer.nvidia.com/rdp/cudnn-download\" target=\"_blank\" rel=\"noopener\">https://developer.nvidia.com/rdp/cudnn-download</a></strong> 下载 <code>cuDNN</code> ，版本为 <code>cudnn-7.0-linux-x64-v3.0-prod.tgz</code></p>\n</li>\n<li><p>安装</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">tar -zxvf cudnn-7.0-linux-x64-v3.0-prod.tgz</span><br><span class=\"line\">cd cuda</span><br><span class=\"line\">sudo cp lib64/* /usr/local/cuda/lib64/</span><br><span class=\"line\">sudo cp include/cudnn.h /usr/local/cuda/include/</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>更新软链接</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cd /usr/local/cuda/lib64</span><br><span class=\"line\">sudo rm -rf libcudnn.so libcudnn.so.7.0</span><br><span class=\"line\">sudo ln -s libcudnn.so.7.0.64 libcudnn.so.7.0</span><br><span class=\"line\">sudo ln -s libcudnn.so.7.0 libcudnn.so</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n<h4 id=\"设置-CUDA-环境变量\"><a href=\"#设置-CUDA-环境变量\" class=\"headerlink\" title=\"设置 CUDA 环境变量\"></a>设置 CUDA 环境变量</h4><ul>\n<li><p><strong>VERSION1</strong></p>\n<ul>\n<li><p>在 <code>/etc/profile</code> 中添加 <code>CUDA</code> 环境变量</p>\n</li>\n<li><p>执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo vim /etc/profile</span><br></pre></td></tr></table></figure>\n<p>在打开的文件中加入如下两句话</p>\n<figure class=\"highlight bash\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">export</span> PATH=/usr/<span class=\"built_in\">local</span>/cuda/bin:<span class=\"variable\">$PATH</span></span><br><span class=\"line\"><span class=\"built_in\">export</span> LD_LIBRARY_PATH=/usr/<span class=\"built_in\">local</span>/cuda/lib64:<span class=\"variable\">$LD_LIBRARY_PATH</span></span><br></pre></td></tr></table></figure>\n</li>\n<li><p>保存后，使环境变量立即生效，执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">source /etc/profile</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n</li>\n<li><p><strong>VERSION2</strong></p>\n<ul>\n<li><p>在 <code>~/.bashrc</code> 中添加 <code>CUDA</code> 环境变量</p>\n</li>\n<li><p>执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo vim ~/.bashrc</span><br></pre></td></tr></table></figure>\n<p>在打开的文件中加入如下两句话</p>\n<figure class=\"highlight bash\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">export</span> PATH=/usr/<span class=\"built_in\">local</span>/cuda/bin:<span class=\"variable\">$PATH</span></span><br><span class=\"line\"><span class=\"built_in\">export</span> LD_LIBRARY_PATH=/usr/<span class=\"built_in\">local</span>/cuda/lib64:<span class=\"variable\">$LD_LIBRARY_PATH</span></span><br></pre></td></tr></table></figure>\n</li>\n<li><p>保存后，使环境变量立即生效，执行</p>\n</li>\n</ul>\n</li>\n<li><p>我用的 <code>VERSION2</code></p>\n</li>\n</ul>\n<h4 id=\"安装cuda-samples\"><a href=\"#安装cuda-samples\" class=\"headerlink\" title=\"安装cuda samples\"></a>安装cuda samples</h4><ul>\n<li><p>进入 <code>/usr/local/cuda/samples</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cd /usr/local/cuda/samples</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>执行下面的命令来 <code>build samples</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo make all -j4</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>全部编译完成后，进入 <code>samples/bin/x86_64/linux/release</code> ，运行 <code>deviceQuery</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cd /usr/local/cuda/samples/bin/x86_64/linux/release</span><br><span class=\"line\">./deviceQuery</span><br></pre></td></tr></table></figure>\n<p>如果出现显卡信息，则驱动及显卡安装成功，结果如下</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">./deviceQuery Starting...</span><br><span class=\"line\"></span><br><span class=\"line\"> CUDA Device Query (Runtime API) version (CUDART static linking)</span><br><span class=\"line\"></span><br><span class=\"line\">Detected 4 CUDA Capable device(s)</span><br><span class=\"line\"></span><br><span class=\"line\">Device 0: \"GeForce GTX 1080 Ti\"</span><br><span class=\"line\">\t.</span><br><span class=\"line\">\t.</span><br><span class=\"line\">\t.</span><br><span class=\"line\">Device 3: \"GeForce GTX 1080 Ti\"</span><br><span class=\"line\"> CUDA Driver Version / Runtime Version 9.0 / 8.0</span><br><span class=\"line\"> CUDA Capability Major/Minor version number: 6.1</span><br><span class=\"line\"> Total amount of global memory: 11172 MBytes (11715084288 bytes)</span><br><span class=\"line\"> (28) Multiprocessors, (128) CUDA Cores/MP: 3584 CUDA Cores</span><br><span class=\"line\"> GPU Max Clock rate: 1582 MHz (1.58 GHz)</span><br><span class=\"line\"> Memory Clock rate: 5505 Mhz</span><br><span class=\"line\"> Memory Bus Width: 352-bit</span><br><span class=\"line\"> L2 Cache Size: 2883584 bytes</span><br><span class=\"line\"> Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)</span><br><span class=\"line\"> Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers</span><br><span class=\"line\"> Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers</span><br><span class=\"line\"> Total amount of constant memory: 65536 bytes</span><br><span class=\"line\"> Total amount of shared memory per block: 49152 bytes</span><br><span class=\"line\"> Total number of registers available per block: 65536</span><br><span class=\"line\"> Warp size: 32</span><br><span class=\"line\"> Maximum number of threads per multiprocessor: 2048</span><br><span class=\"line\"> Maximum number of threads per block: 1024</span><br><span class=\"line\"> Max dimension size of a thread block (x,y,z): (1024, 1024, 64)</span><br><span class=\"line\"> Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)</span><br><span class=\"line\"> Maximum memory pitch: 2147483647 bytes</span><br><span class=\"line\"> Texture alignment: 512 bytes</span><br><span class=\"line\"> Concurrent copy and kernel execution: Yes with 2 copy engine(s)</span><br><span class=\"line\"> Run time limit on kernels: No</span><br><span class=\"line\"> Integrated GPU sharing Host Memory: No</span><br><span class=\"line\"> Support host page-locked memory mapping: Yes</span><br><span class=\"line\"> Alignment requirement for Surfaces: Yes</span><br><span class=\"line\"> Device has ECC support: Disabled</span><br><span class=\"line\"> Device supports Unified Addressing (UVA): Yes</span><br><span class=\"line\"> Device PCI Domain ID / Bus ID / location ID: 0 / 131 / 0</span><br><span class=\"line\"> Compute Mode:</span><br><span class=\"line\"> < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) ></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU0) -> GeForce GTX 1080 Ti (GPU1) : Yes</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU0) -> GeForce GTX 1080 Ti (GPU2) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU0) -> GeForce GTX 1080 Ti (GPU3) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU1) -> GeForce GTX 1080 Ti (GPU0) : Yes</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU1) -> GeForce GTX 1080 Ti (GPU2) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU1) -> GeForce GTX 1080 Ti (GPU3) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU2) -> GeForce GTX 1080 Ti (GPU0) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU2) -> GeForce GTX 1080 Ti (GPU1) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU2) -> GeForce GTX 1080 Ti (GPU3) : Yes</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU3) -> GeForce GTX 1080 Ti (GPU0) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU3) -> GeForce GTX 1080 Ti (GPU1) : No</span></span><br><span class=\"line\"><span class=\"meta\">></span><span class=\"bash\"> Peer access from GeForce GTX 1080 Ti (GPU3) -> GeForce GTX 1080 Ti (GPU2) : Yes</span></span><br><span class=\"line\"></span><br><span class=\"line\">deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 9.0, CUDA Runtime Version = 8.0, NumDevs = 4, Device0 = GeForce GTX 1080 Ti, Device1 = GeForce GTX 1080 Ti, Device2 = GeForce GTX 1080 Ti, Device3 = GeForce GTX 1080 Ti</span><br><span class=\"line\">Result = PASS</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>因为我是 4 个 GPU ，所以会有多个 Device 信息</p>\n</blockquote>\n<p>安装成功！</p>\n</li>\n</ul>\n<h4 id=\"安装Intel-MKL，openBlas-或Atlas\"><a href=\"#安装Intel-MKL，openBlas-或Atlas\" class=\"headerlink\" title=\"安装Intel MKL，openBlas 或Atlas\"></a>安装Intel MKL，openBlas 或Atlas</h4><p>我选择的是 <code>Atlas</code> ，为 <code>caffe</code> 默认使用的，不要额外配置，安装命令</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo apt-get install libatlas-base-dev</span><br></pre></td></tr></table></figure>\n<h4 id=\"安装opencv\"><a href=\"#安装opencv\" class=\"headerlink\" title=\"安装opencv\"></a>安装opencv</h4><ul>\n<li><p>首先安装必须的包</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>从 <code>opencv</code> 官网下载最新版本 <code>opencv</code> ，<code>3.2.0.zip</code></p>\n<p>解压</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">unzip 3.2.0.zip</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>编译</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cd opencv-3.2.0</span><br><span class=\"line\">mkdir release</span><br><span class=\"line\">cd release</span><br><span class=\"line\">cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local .. -DWITH_LAPACK=OFF</span><br><span class=\"line\">make -j4</span><br><span class=\"line\">sudo make install</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>测试</p>\n<p>因为我的 <code>GPU</code> 服务器走的别人的网关，这里显示不出来图片，也就没有测试。</p>\n<p>但是是需要测试的，自行 <code>Google</code> 吧</p>\n</li>\n</ul>\n<h4 id=\"安装python依赖库\"><a href=\"#安装python依赖库\" class=\"headerlink\" title=\"安装python依赖库\"></a>安装python依赖库</h4><ul>\n<li><p>下载 <code>caffe</code> 源码，解压后进入 <code>caffe-master</code> 下的 <code>python</code> 目录</p>\n<blockquote>\n<p>这里一定要注意当前下载的 caffe 版本是否支持自己安装的 cuDNN 版本</p>\n</blockquote>\n</li>\n<li><p>安装 <code>python-pip</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">sudo apt-get install python-pip</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>执行如下命令安装相关依赖</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">for req in $(cat requirements.txt); do pip install $req; done</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n<h4 id=\"编译Caffe\"><a href=\"#编译Caffe\" class=\"headerlink\" title=\"编译Caffe\"></a>编译Caffe</h4><ul>\n<li><p>进入 <code>caffe-master</code> 目录，复制一份 <code>Makefile.config.examples</code> </p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cp Makefile.config.example Makefile.config</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>修改 <code>Makefile.config</code></p>\n<p>开启对 <code>CuDNN</code> 和 <code>Opencv3</code> 的支持</p>\n<figure class=\"highlight makefile\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"comment\">## Refer to http://caffe.berkeleyvision.org/installation.html</span></span><br><span class=\"line\"><span class=\"comment\"># Contributions simplifying and improving our build system are welcome!</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># cuDNN acceleration switch (uncomment to build with cuDNN).</span></span><br><span class=\"line\">USE_CUDNN := 1</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># CPU-only switch (uncomment to build without GPU support).</span></span><br><span class=\"line\"><span class=\"comment\"># CPU_ONLY := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># uncomment to disable IO dependencies and corresponding data layers</span></span><br><span class=\"line\"><span class=\"comment\"># USE_OPENCV := 0</span></span><br><span class=\"line\"><span class=\"comment\"># USE_LEVELDB := 0</span></span><br><span class=\"line\"><span class=\"comment\"># USE_LMDB := 0</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)</span></span><br><span class=\"line\"><span class=\"comment\">#\tYou should not set this flag if you will be reading LMDBs with any</span></span><br><span class=\"line\"><span class=\"comment\">#\tpossibility of simultaneous read and write</span></span><br><span class=\"line\"><span class=\"comment\"># ALLOW_LMDB_NOLOCK := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment if you're using OpenCV 3</span></span><br><span class=\"line\">OPENCV_VERSION := 3</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># To customize your choice of compiler, uncomment and set the following.</span></span><br><span class=\"line\"><span class=\"comment\"># N.B. the default for Linux is g++ and the default for OSX is clang++</span></span><br><span class=\"line\"><span class=\"comment\"># CUSTOM_CXX := g++</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># CUDA directory contains bin/ and lib/ directories that we need.</span></span><br><span class=\"line\">CUDA_DIR := /usr/local/cuda</span><br><span class=\"line\"><span class=\"comment\"># On Ubuntu 14.04, if cuda tools are installed via</span></span><br><span class=\"line\"><span class=\"comment\"># \"sudo apt-get install nvidia-cuda-toolkit\" then use this instead:</span></span><br><span class=\"line\"><span class=\"comment\"># CUDA_DIR := /usr</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># CUDA architecture setting: going with all of them.</span></span><br><span class=\"line\"><span class=\"comment\"># For CUDA < 6.0, comment the *_50 lines for compatibility.</span></span><br><span class=\"line\">CUDA_ARCH := -gencode arch=compute_20,code=sm_20 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_20,code=sm_21 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_30,code=sm_30 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_35,code=sm_35 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_50,code=sm_50 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_50,code=compute_50</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># BLAS choice:</span></span><br><span class=\"line\"><span class=\"comment\"># atlas for ATLAS (default)</span></span><br><span class=\"line\"><span class=\"comment\"># mkl for MKL</span></span><br><span class=\"line\"><span class=\"comment\"># open for OpenBlas</span></span><br><span class=\"line\">BLAS := atlas</span><br><span class=\"line\"><span class=\"comment\"># Custom (MKL/ATLAS/OpenBLAS) include and lib directories.</span></span><br><span class=\"line\"><span class=\"comment\"># Leave commented to accept the defaults for your choice of BLAS</span></span><br><span class=\"line\"><span class=\"comment\"># (which should work)!</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_INCLUDE := /path/to/your/blas</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_LIB := /path/to/your/blas</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Homebrew puts openblas in a directory that is not on the standard search path</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_INCLUDE := $(shell brew --prefix openblas)/include</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_LIB := $(shell brew --prefix openblas)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># This is required only if you will compile the matlab interface.</span></span><br><span class=\"line\"><span class=\"comment\"># MATLAB directory should contain the mex binary in /bin.</span></span><br><span class=\"line\"><span class=\"comment\"># MATLAB_DIR := /usr/local</span></span><br><span class=\"line\"><span class=\"comment\"># MATLAB_DIR := /Applications/MATLAB_R2012b.app</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># <span class=\"doctag\">NOTE:</span> this is required only if you will compile the python interface.</span></span><br><span class=\"line\"><span class=\"comment\"># We need to be able to find Python.h and numpy/arrayobject.h.</span></span><br><span class=\"line\">PYTHON_INCLUDE := /usr/<span class=\"keyword\">include</span>/python2.7 \\</span><br><span class=\"line\">\t\t/usr/lib/python2.7/dist-packages/numpy/core/<span class=\"keyword\">include</span></span><br><span class=\"line\"><span class=\"comment\"># Anaconda Python distribution is quite popular. Include path:</span></span><br><span class=\"line\"><span class=\"comment\"># Verify anaconda location, sometimes it's in root.</span></span><br><span class=\"line\"><span class=\"comment\"># ANACONDA_HOME := $(HOME)/anaconda</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_INCLUDE := $(ANACONDA_HOME)/include \\</span></span><br><span class=\"line\">\t\t<span class=\"comment\"># $(ANACONDA_HOME)/include/python2.7 \\</span></span><br><span class=\"line\">\t\t<span class=\"comment\"># $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \\</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment to use Python 3 (default is Python 2)</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_LIBRARIES := boost_python3 python3.5m</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_INCLUDE := /usr/include/python3.5m \\</span></span><br><span class=\"line\"><span class=\"comment\"># /usr/lib/python3.5/dist-packages/numpy/core/include</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># We need to be able to find libpythonX.X.so or .dylib.</span></span><br><span class=\"line\">PYTHON_LIB := /usr/lib</span><br><span class=\"line\"><span class=\"comment\"># PYTHON_LIB := $(ANACONDA_HOME)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Homebrew installs numpy in a non standard path (keg only)</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_LIB += $(shell brew --prefix numpy)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment to support layers written in Python (will link against Python libs)</span></span><br><span class=\"line\"><span class=\"comment\"># WITH_PYTHON_LAYER := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Whatever else you find you need goes here.</span></span><br><span class=\"line\">INCLUDE_DIRS := <span class=\"variable\">$(PYTHON_INCLUDE)</span> /usr/local/<span class=\"keyword\">include</span></span><br><span class=\"line\">LIBRARY_DIRS := <span class=\"variable\">$(PYTHON_LIB)</span> /usr/local/lib /usr/lib</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies</span></span><br><span class=\"line\"><span class=\"comment\"># INCLUDE_DIRS += $(shell brew --prefix)/include</span></span><br><span class=\"line\"><span class=\"comment\"># LIBRARY_DIRS += $(shell brew --prefix)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment to use `pkg-config` to specify OpenCV library paths.</span></span><br><span class=\"line\"><span class=\"comment\"># (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)</span></span><br><span class=\"line\"><span class=\"comment\"># USE_PKG_CONFIG := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\">BUILD_DIR := build</span><br><span class=\"line\">DISTRIBUTE_DIR := distribute</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171</span></span><br><span class=\"line\"><span class=\"comment\"># DEBUG := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># The ID of the GPU that 'make runtest' will use to run unit tests.</span></span><br><span class=\"line\">TEST_GPUID := 0</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># enable pretty build (comment to see full commands)</span></span><br><span class=\"line\">Q ?= @</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>保存退出，编译</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">make all -j4</span><br><span class=\"line\">make test -j4</span><br><span class=\"line\">make runtest -j4</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>结果</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">[ FAILED ] 5 tests, listed below:</span><br><span class=\"line\">[ FAILED ] CuDNNConvolutionLayerTest/0.TestSimpleConvolutionCuDNN, where TypeParam = float</span><br><span class=\"line\">[ FAILED ] CuDNNConvolutionLayerTest/0.TestGradientCuDNN, where TypeParam = float</span><br><span class=\"line\">[ FAILED ] CuDNNConvolutionLayerTest/0.TestGradientGroupCuDNN, where TypeParam = float</span><br><span class=\"line\">[ FAILED ] CuDNNConvolutionLayerTest/0.TestSobelConvolutionCuDNN, where TypeParam = float</span><br><span class=\"line\">[ FAILED ] CuDNNConvolutionLayerTest/0.TestSimpleConvolutionGroupCuDNN, where TypeParam = float</span><br></pre></td></tr></table></figure>\n<p>这个是因为我们使用的是 <code>cuDNN v3</code> 版本，而这部分只在 <code>cuDNN v4</code> 下 <code>work</code></p>\n</li>\n</ul>\n<h3 id=\"更换cuDNN与caffe版本\"><a href=\"#更换cuDNN与caffe版本\" class=\"headerlink\" title=\"更换cuDNN与caffe版本\"></a>更换cuDNN与caffe版本</h3><h4 id=\"为啥要换\"><a href=\"#为啥要换\" class=\"headerlink\" title=\"为啥要换\"></a>为啥要换</h4><p>我当然不想换！因为我老大不太满意！也没有说不满意，就是素质三连了一下</p>\n<p>都行</p>\n<p>随你</p>\n<p>都可以</p>\n<p>。。。</p>\n<h4 id=\"确定依赖关系\"><a href=\"#确定依赖关系\" class=\"headerlink\" title=\"确定依赖关系\"></a>确定依赖关系</h4><ul>\n<li><p><code>CuDNN</code> 是专门针对 <code>Deep Learning</code> 框架设计的一套 <code>GPU</code> 计算加速方案，目前支持的 <code>DL</code> 库包括 <code>Caffe</code> ，<code>ConvNet</code> ，<code>Torch7</code> 等</p>\n<p>基本原理是把 <code>lib</code> 文件加入到系统能找到的 <code>lib</code> 文件夹里， 把头文件加到系统能找到的 <code>include</code> 文件夹里就可以。这里把他们加到 <code>CUDA</code> 的文件夹下</p>\n</li>\n<li><p><code>CUDA</code> 是直接安装的、<code>OpenCV</code> 安装的时候是依赖于 <code>CUDA</code> 的、<code>cuDNN</code> 的安装就是将其 <code>lib文件加入到系统能找到的lib文件夹里， 把头文件加到系统能找到的include文件夹里，这里把他们加到CUDA的文件夹下</code></p>\n</li>\n<li><p>所以说 <code>cuDNN</code> 无关于 <code>CUDA</code> ，无关于 <code>OpenCV</code> ，它只和 <code>caffe</code> 有关系，因为 <code>caffe</code> 安装的时候要依赖它。</p>\n</li>\n</ul>\n<h4 id=\"重装过程\"><a href=\"#重装过程\" class=\"headerlink\" title=\"重装过程\"></a>重装过程</h4><ul>\n<li><p><code>cuDNN</code> 文件结构</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">root@Jedy:/home/jedy/deeplearning/cuda# tree -N</span><br><span class=\"line\">.</span><br><span class=\"line\">├── include</span><br><span class=\"line\">│ └── cudnn.h</span><br><span class=\"line\">└── lib64</span><br><span class=\"line\"> ├── libcudnn.so</span><br><span class=\"line\"> ├── libcudnn.so.5</span><br><span class=\"line\"> ├── libcudnn.so.5.0.5</span><br><span class=\"line\"> └── libcudnn_static.a</span><br><span class=\"line\"></span><br><span class=\"line\">2 directories, 5 files</span><br></pre></td></tr></table></figure>\n<p>而安装 <code>cuDNN</code> 的命令是</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">tar -zxvf cudnn-7.0-linux-x64-v3.0-prod.tgz </span><br><span class=\"line\">cd cuda</span><br><span class=\"line\">sudo cp lib64/* /usr/local/cuda/lib64/</span><br><span class=\"line\">sudo cp include/cudnn.h /usr/local/cuda/include/</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\">更新软链接</span></span><br><span class=\"line\">cd /usr/local/cuda/lib64</span><br><span class=\"line\">sudo rm -rf libcudnn.so libcudnn.so.7.0</span><br><span class=\"line\">sudo ln -s libcudnn.so.7.0.64 libcudnn.so.7.0</span><br><span class=\"line\">sudo ln -s libcudnn.so.7.0 libcudnn.so</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>重装 <code>cuDNN</code> </p>\n<ul>\n<li><p>因为 <code>cuDNN</code> 只是将文件拷贝进去，所以我们只需要下载更新版本的 <code>cuDNN</code> 然后覆盖安装即可。我是打算一步到位，使用最新的 <code>caffe</code> ，又怕 <code>cuDNN</code> 向下兼容不好，所以下载了 <code>v5</code> 版本，版本为 <code>cudnn-8.0-linux-x64-v5.0-ga.tgz</code></p>\n</li>\n<li><p>进入到 <code>/usr/local/cuda/lib64</code> 中，删除旧版本的 <code>cuDNN</code></p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cd /usr/local/cuda/lib64</span><br><span class=\"line\">sudo rm -rf libcudnn.so libcudnn.so.7.0 libcudnn.so.7.0.64</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>解压新版本的 <code>cuDNN</code> 并拷贝</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">tar -zxvf cudnn-8.0-linux-x64-v5.0-ga.tgz</span><br><span class=\"line\">cd cuda</span><br><span class=\"line\">sudo cp lib64/* /usr/local/cuda/lib64/</span><br><span class=\"line\">sudo cp include/cudnn.h /usr/local/cuda/include/</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>这里的 include/cudnn.h 我们并没有删除旧版本的，而是采用了覆盖安装的方式</p>\n</blockquote>\n</li>\n<li><p>更新软链接</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cd /usr/local/cuda/lib64</span><br><span class=\"line\">sudo rm -rf libcudnn.so libcudnn.so.5</span><br><span class=\"line\">sudo ln -s libcudnn.so.5.0.5 libcudnn.so.5</span><br><span class=\"line\">sudo ln -s libcudnn.so.5 libcudnn.so</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n</li>\n<li><p>重装 <code>caffe</code></p>\n<ul>\n<li><p>在 <code>caffe</code> 的 <code>root</code> 目录执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">make clean</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>删除旧版本的 <code>caffe</code> </p>\n</li>\n<li><p>下载新版本的 <code>caffe</code> ，我这里直接下载了 <code>1.0</code> 版本的 <code>caffe</code> </p>\n</li>\n<li><p>进入 <code>caffe-master</code> 目录，复制一份 <code>Makefile.config.examples</code> </p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">cp Makefile.config.example Makefile.config</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>修改 <code>Makefile.config</code> ，开启 <code>CuDNN</code> 和 <code>Opencv3</code> 的支持即可，修改好的如下</p>\n<figure class=\"highlight makefile\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"comment\">## Refer to http://caffe.berkeleyvision.org/installation.html</span></span><br><span class=\"line\"><span class=\"comment\"># Contributions simplifying and improving our build system are welcome!</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># cuDNN acceleration switch (uncomment to build with cuDNN).</span></span><br><span class=\"line\">USE_CUDNN := 1</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># CPU-only switch (uncomment to build without GPU support).</span></span><br><span class=\"line\"><span class=\"comment\"># CPU_ONLY := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># uncomment to disable IO dependencies and corresponding data layers</span></span><br><span class=\"line\"><span class=\"comment\"># USE_OPENCV := 0</span></span><br><span class=\"line\"><span class=\"comment\"># USE_LEVELDB := 0</span></span><br><span class=\"line\"><span class=\"comment\"># USE_LMDB := 0</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># uncomment to allow MDB_NOLOCK when reading LMDB files (only if necessary)</span></span><br><span class=\"line\"><span class=\"comment\">#\tYou should not set this flag if you will be reading LMDBs with any</span></span><br><span class=\"line\"><span class=\"comment\">#\tpossibility of simultaneous read and write</span></span><br><span class=\"line\"><span class=\"comment\"># ALLOW_LMDB_NOLOCK := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment if you're using OpenCV 3</span></span><br><span class=\"line\">OPENCV_VERSION := 3</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># To customize your choice of compiler, uncomment and set the following.</span></span><br><span class=\"line\"><span class=\"comment\"># N.B. the default for Linux is g++ and the default for OSX is clang++</span></span><br><span class=\"line\"><span class=\"comment\"># CUSTOM_CXX := g++</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># CUDA directory contains bin/ and lib/ directories that we need.</span></span><br><span class=\"line\">CUDA_DIR := /usr/local/cuda</span><br><span class=\"line\"><span class=\"comment\"># On Ubuntu 14.04, if cuda tools are installed via</span></span><br><span class=\"line\"><span class=\"comment\"># \"sudo apt-get install nvidia-cuda-toolkit\" then use this instead:</span></span><br><span class=\"line\"><span class=\"comment\"># CUDA_DIR := /usr</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># CUDA architecture setting: going with all of them.</span></span><br><span class=\"line\"><span class=\"comment\"># For CUDA < 6.0, comment the *_50 through *_61 lines for compatibility.</span></span><br><span class=\"line\"><span class=\"comment\"># For CUDA < 8.0, comment the *_60 and *_61 lines for compatibility.</span></span><br><span class=\"line\">CUDA_ARCH := -gencode arch=compute_20,code=sm_20 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_20,code=sm_21 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_30,code=sm_30 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_35,code=sm_35 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_50,code=sm_50 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_52,code=sm_52 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_60,code=sm_60 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_61,code=sm_61 \\</span><br><span class=\"line\">\t\t-gencode arch=compute_61,code=compute_61</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># BLAS choice:</span></span><br><span class=\"line\"><span class=\"comment\"># atlas for ATLAS (default)</span></span><br><span class=\"line\"><span class=\"comment\"># mkl for MKL</span></span><br><span class=\"line\"><span class=\"comment\"># open for OpenBlas</span></span><br><span class=\"line\">BLAS := atlas</span><br><span class=\"line\"><span class=\"comment\"># Custom (MKL/ATLAS/OpenBLAS) include and lib directories.</span></span><br><span class=\"line\"><span class=\"comment\"># Leave commented to accept the defaults for your choice of BLAS</span></span><br><span class=\"line\"><span class=\"comment\"># (which should work)!</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_INCLUDE := /path/to/your/blas</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_LIB := /path/to/your/blas</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Homebrew puts openblas in a directory that is not on the standard search path</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_INCLUDE := $(shell brew --prefix openblas)/include</span></span><br><span class=\"line\"><span class=\"comment\"># BLAS_LIB := $(shell brew --prefix openblas)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># This is required only if you will compile the matlab interface.</span></span><br><span class=\"line\"><span class=\"comment\"># MATLAB directory should contain the mex binary in /bin.</span></span><br><span class=\"line\"><span class=\"comment\"># MATLAB_DIR := /usr/local</span></span><br><span class=\"line\"><span class=\"comment\"># MATLAB_DIR := /Applications/MATLAB_R2012b.app</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># <span class=\"doctag\">NOTE:</span> this is required only if you will compile the python interface.</span></span><br><span class=\"line\"><span class=\"comment\"># We need to be able to find Python.h and numpy/arrayobject.h.</span></span><br><span class=\"line\">PYTHON_INCLUDE := /usr/<span class=\"keyword\">include</span>/python2.7 \\</span><br><span class=\"line\">\t\t/usr/lib/python2.7/dist-packages/numpy/core/<span class=\"keyword\">include</span></span><br><span class=\"line\"><span class=\"comment\"># Anaconda Python distribution is quite popular. Include path:</span></span><br><span class=\"line\"><span class=\"comment\"># Verify anaconda location, sometimes it's in root.</span></span><br><span class=\"line\"><span class=\"comment\"># ANACONDA_HOME := $(HOME)/anaconda</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_INCLUDE := $(ANACONDA_HOME)/include \\</span></span><br><span class=\"line\">\t\t<span class=\"comment\"># $(ANACONDA_HOME)/include/python2.7 \\</span></span><br><span class=\"line\">\t\t<span class=\"comment\"># $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment to use Python 3 (default is Python 2)</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_LIBRARIES := boost_python3 python3.5m</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_INCLUDE := /usr/include/python3.5m \\</span></span><br><span class=\"line\"><span class=\"comment\"># /usr/lib/python3.5/dist-packages/numpy/core/include</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># We need to be able to find libpythonX.X.so or .dylib.</span></span><br><span class=\"line\">PYTHON_LIB := /usr/lib</span><br><span class=\"line\"><span class=\"comment\"># PYTHON_LIB := $(ANACONDA_HOME)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Homebrew installs numpy in a non standard path (keg only)</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_INCLUDE += $(dir $(shell python -c 'import numpy.core; print(numpy.core.__file__)'))/include</span></span><br><span class=\"line\"><span class=\"comment\"># PYTHON_LIB += $(shell brew --prefix numpy)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment to support layers written in Python (will link against Python libs)</span></span><br><span class=\"line\"><span class=\"comment\"># WITH_PYTHON_LAYER := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Whatever else you find you need goes here.</span></span><br><span class=\"line\">INCLUDE_DIRS := <span class=\"variable\">$(PYTHON_INCLUDE)</span> /usr/local/<span class=\"keyword\">include</span></span><br><span class=\"line\">LIBRARY_DIRS := <span class=\"variable\">$(PYTHON_LIB)</span> /usr/local/lib /usr/lib</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># If Homebrew is installed at a non standard location (for example your home directory) and you use it for general dependencies</span></span><br><span class=\"line\"><span class=\"comment\"># INCLUDE_DIRS += $(shell brew --prefix)/include</span></span><br><span class=\"line\"><span class=\"comment\"># LIBRARY_DIRS += $(shell brew --prefix)/lib</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># NCCL acceleration switch (uncomment to build with NCCL)</span></span><br><span class=\"line\"><span class=\"comment\"># https://github.com/NVIDIA/nccl (last tested version: v1.2.3-1+cuda8.0)</span></span><br><span class=\"line\"><span class=\"comment\"># USE_NCCL := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment to use `pkg-config` to specify OpenCV library paths.</span></span><br><span class=\"line\"><span class=\"comment\"># (Usually not necessary -- OpenCV libraries are normally installed in one of the above $LIBRARY_DIRS.)</span></span><br><span class=\"line\"><span class=\"comment\"># USE_PKG_CONFIG := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># N.B. both build and distribute dirs are cleared on `make clean`</span></span><br><span class=\"line\">BUILD_DIR := build</span><br><span class=\"line\">DISTRIBUTE_DIR := distribute</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># Uncomment for debugging. Does not work on OSX due to https://github.com/BVLC/caffe/issues/171</span></span><br><span class=\"line\"><span class=\"comment\"># DEBUG := 1</span></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># The ID of the GPU that 'make runtest' will use to run unit tests.</span></span><br><span class=\"line\">TEST_GPUID := 0</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"comment\"># enable pretty build (comment to see full commands)</span></span><br><span class=\"line\">Q ?= @</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>保存退出，编译</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">make all -j4</span><br><span class=\"line\">make test -j4</span><br><span class=\"line\">make runtest -j4</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>没有报错，成功</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">[----------] Global test environment tear-down</span><br><span class=\"line\">[==========] 2101 tests from 277 test cases ran. (371248 ms total)</span><br><span class=\"line\">[ PASSED ] 2101 tests.</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n</li>\n</ul>\n<h4 id=\"修改后的组件版本\"><a href=\"#修改后的组件版本\" class=\"headerlink\" title=\"修改后的组件版本\"></a>修改后的组件版本</h4><ul>\n<li><strong>Ubuntu</strong><ul>\n<li>14.04</li>\n</ul>\n</li>\n</ul>\n<ul>\n<li><strong>CUDA</strong><ul>\n<li>8.0</li>\n</ul>\n</li>\n<li><strong>cuDNN</strong><ul>\n<li>cudnn-8.0-linux-x64-v5.0-ga.tgz</li>\n</ul>\n</li>\n<li><strong>opencv</strong><ul>\n<li>3.2</li>\n</ul>\n</li>\n<li><strong>python</strong><ul>\n<li>Python 2.7.6</li>\n</ul>\n</li>\n<li><strong>caffe</strong><ul>\n<li>1.0</li>\n</ul>\n</li>\n</ul>\n<h3 id=\"一点心得\"><a href=\"#一点心得\" class=\"headerlink\" title=\"一点心得\"></a>一点心得</h3><ul>\n<li><p>make 编译的时候可以通过</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">make -j4</span><br></pre></td></tr></table></figure>\n<p>指令指定多线程编译，可以很大程度加快编译速度。具体可以开多少看自己机器配置</p>\n</li>\n<li><p>看好OpenCV支持的CUDA版本</p>\n</li>\n<li><p>看好caffe支持的cuDNN版本</p>\n</li>\n<li><p>每个人硬件、环境的不同都会导致踩到不同的坑。这个就多 Google 吧！</p>\n</li>\n</ul>\n<blockquote>\n<p>部分参考 <a href=\"http://www.cnblogs.com/wm123/p/5385940.html\" target=\"_blank\" rel=\"noopener\"><strong>Ubuntu 14.04上安装caffe</strong></a></p>\n</blockquote>\n","categories":["安装记录"],"tags":["linux","机器学习","caffe"]},{"title":"git 使用小结","url":"https://www.holmeyoung.com/blog/git-summary/","content":"<h2 id=\"新建-GitLab-仓库时官方给出的命令\"><a href=\"#新建-GitLab-仓库时官方给出的命令\" class=\"headerlink\" title=\"新建 GitLab 仓库时官方给出的命令\"></a>新建 GitLab 仓库时官方给出的命令</h2><h3 id=\"Git-global-setup\"><a href=\"#Git-global-setup\" class=\"headerlink\" title=\"Git global setup\"></a><strong>Git global setup</strong></h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git config --global user.name <span class=\"string\">\"YourName\"</span></span><br><span class=\"line\">git config --global user.email <span class=\"string\">\"YourEmail\"</span></span><br></pre></td></tr></table></figure>\n<h3 id=\"Create-a-new-repository\"><a href=\"#Create-a-new-repository\" class=\"headerlink\" title=\"Create a new repository\"></a>Create a new repository</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git <span class=\"built_in\">clone</span> https://gitlab.com/holmeyoung/num_recog.git</span><br><span class=\"line\"><span class=\"built_in\">cd</span> num_recog</span><br><span class=\"line\">touch README.md</span><br><span class=\"line\">git add README.md</span><br><span class=\"line\">git commit -m <span class=\"string\">\"add README\"</span></span><br><span class=\"line\">git push -u origin master</span><br></pre></td></tr></table></figure>\n<h3 id=\"Existing-folder\"><a href=\"#Existing-folder\" class=\"headerlink\" title=\"Existing folder\"></a><strong>Existing folder</strong></h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">cd</span> existing_folder</span><br><span class=\"line\">git init</span><br><span class=\"line\">git remote add origin https://gitlab.com/holmeyoung/num_recog.git</span><br><span class=\"line\">git add .</span><br><span class=\"line\">git commit -m <span class=\"string\">\"Initial commit\"</span></span><br><span class=\"line\">git push -u origin master</span><br></pre></td></tr></table></figure>\n<h3 id=\"Existing-Git-repository\"><a href=\"#Existing-Git-repository\" class=\"headerlink\" title=\"Existing Git repository\"></a><strong>Existing Git repository</strong></h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">cd</span> existing_repo</span><br><span class=\"line\">git remote rename origin old-origin</span><br><span class=\"line\">git remote add origin https://gitlab.com/holmeyoung/num_recog.git</span><br><span class=\"line\">git push -u origin --all</span><br><span class=\"line\">git push -u origin --tags</span><br></pre></td></tr></table></figure>\n<h2 id=\"新建-GitHab-仓库时官方给出的命令\"><a href=\"#新建-GitHab-仓库时官方给出的命令\" class=\"headerlink\" title=\"新建 GitHab 仓库时官方给出的命令\"></a>新建 GitHab 仓库时官方给出的命令</h2><h3 id=\"…or-create-a-new-repository-on-the-command-line\"><a href=\"#…or-create-a-new-repository-on-the-command-line\" class=\"headerlink\" title=\"…or create a new repository on the command line\"></a>…or create a new repository on the command line</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">echo</span> <span class=\"string\">\"# tmp\"</span> >> README.md</span><br><span class=\"line\">git init</span><br><span class=\"line\">git add README.md</span><br><span class=\"line\">git commit -m <span class=\"string\">\"first commit\"</span></span><br><span class=\"line\">git remote add origin [email protected]:Holmeyoung/tmp.git</span><br><span class=\"line\">git push -u origin master</span><br></pre></td></tr></table></figure>\n<h3 id=\"…or-push-an-existing-repository-from-the-command-line\"><a href=\"#…or-push-an-existing-repository-from-the-command-line\" class=\"headerlink\" title=\"…or push an existing repository from the command line\"></a>…or push an existing repository from the command line</h3><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git remote add origin [email protected]:Holmeyoung/tmp.git</span><br><span class=\"line\">git push -u origin master</span><br></pre></td></tr></table></figure>\n<h3 id=\"…or-import-code-from-another-repository\"><a href=\"#…or-import-code-from-another-repository\" class=\"headerlink\" title=\"…or import code from another repository\"></a>…or import code from another repository</h3><p>You can initialize this repository with code from a Subversion, Mercurial, or TFS project.</p>\n<h2 id=\"Git-免密码\"><a href=\"#Git-免密码\" class=\"headerlink\" title=\"Git 免密码\"></a>Git 免密码</h2><h3 id=\"Git-HTTPS-方式传输文件免密码\"><a href=\"#Git-HTTPS-方式传输文件免密码\" class=\"headerlink\" title=\"Git HTTPS 方式传输文件免密码\"></a><strong>Git <code>HTTPS</code> 方式传输文件免密码</strong></h3><h4 id=\"cache\"><a href=\"#cache\" class=\"headerlink\" title=\"cache\"></a><strong>cache</strong></h4><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git config --global credential.helper cache</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>或直接修改 <strong><code>~/.gitconfig</code></strong></p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> [credential]</span><br><span class=\"line\">> helper = cache --timeout=3600</span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>采用 <code>cache</code> 形式，密码会临时保存在 <code>~/git-credential-cache</code> 文件夹下</p>\n</blockquote>\n<h4 id=\"store\"><a href=\"#store\" class=\"headerlink\" title=\"store\"></a><strong>store</strong></h4><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git config --global credential.helper store</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>或直接修改 <strong><code>~/.gitconfig</code></strong></p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">> [credential]</span><br><span class=\"line\">> helper = store</span><br><span class=\"line\">></span><br></pre></td></tr></table></figure>\n</blockquote>\n<blockquote>\n<p>采用 <code>store</code> 形式，密码会永久保存在 <code>~/.git-credentials</code> 文件中</p>\n</blockquote>\n<h3 id=\"Git-SSH-方式传输文件免密码\"><a href=\"#Git-SSH-方式传输文件免密码\" class=\"headerlink\" title=\"Git SSH 方式传输文件免密码\"></a>Git <code>SSH</code> 方式传输文件免密码</h3><p><strong>配置密钥即可</strong></p>\n<h2 id=\"分支管理\"><a href=\"#分支管理\" class=\"headerlink\" title=\"分支管理\"></a>分支管理</h2><h3 id=\"远程服务器（远程仓库）\"><a href=\"#远程服务器（远程仓库）\" class=\"headerlink\" title=\"远程服务器（远程仓库）\"></a>远程服务器（远程仓库）</h3><h4 id=\"远程服务器（远程仓库）是什么\"><a href=\"#远程服务器（远程仓库）是什么\" class=\"headerlink\" title=\"远程服务器（远程仓库）是什么\"></a>远程服务器（远程仓库）是什么</h4><p>有命令<br><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git push origin master:master</span><br></pre></td></tr></table></figure></p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git pull origin master:master</span><br></pre></td></tr></table></figure>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git fetch origin</span><br></pre></td></tr></table></figure>\n<p><strong>这里的 <code>origin</code> 究竟是什么</strong></p>\n<p>有命令<br><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git remote add origin [email protected]:shuly/wiki.git</span><br></pre></td></tr></table></figure></p>\n<blockquote>\n<p>所以 </p>\n<p><code>origin --> [email protected]:shuly/wiki.git</code></p>\n<p><code>origin</code> 即为指向远程服务器 <a href=\"mailto:`[email protected]\" target=\"_blank\" rel=\"noopener\">`[email protected]</a>:shuly/wiki.git` 的远程服务器名</p>\n</blockquote>\n<h4 id=\"多个远程服务器\"><a href=\"#多个远程服务器\" class=\"headerlink\" title=\"多个远程服务器\"></a>多个远程服务器</h4><p>假设还有另一个服务器 <a href=\"mailto:`[email protected]\" target=\"_blank\" rel=\"noopener\">`[email protected]</a><code>。可以用第二章中提到的</code>git remote add<code>命令把它加为当前项目的远程分支之一。我们把它命名为</code>origin_dev` ，以便代替完整的 Git URL 以方便使用。</p>\n<p>执行命令<br><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git remote add origin_dev [email protected]:Holmeyoung/wiki.git</span><br></pre></td></tr></table></figure></p>\n<p>现在你可以用 <code>git fetch origin_dev</code> 来获取另一台服务器上你还没有的数据了。执行该命令会创建一个名为 <code>origin_dev/master</code> 的远程分支，指向 <code>origin_dev</code> 服务器上 <code>master</code> 分支所在的提交对象</p>\n<h3 id=\"远程分支\"><a href=\"#远程分支\" class=\"headerlink\" title=\"远程分支\"></a>远程分支</h3><h4 id=\"删除远程分支\"><a href=\"#删除远程分支\" class=\"headerlink\" title=\"删除远程分支\"></a>删除远程分支</h4><p>如果不再需要某个远程分支了，比如搞定了某个特性并把它合并进了远程的 <code>master</code> 分支（或任何其他存放稳定代码的分支），可以用这个非常无厘头的语法来删除它：<code>git push [远程名] :[分支名]</code> 。如果想在服务器上删除 develop 分支，运行下面的命令：<br><figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git push origin :develop</span><br></pre></td></tr></table></figure></p>\n<h3 id=\"本地分支\"><a href=\"#本地分支\" class=\"headerlink\" title=\"本地分支\"></a>本地分支</h3><ul>\n<li><p>创建一个叫做 <code>feature_x</code> 的分支，并切换过去：</p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git checkout -b feature_x</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>切换回主分支：</p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git checkout master</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>再把新建的分支删掉：</p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git branch -d feature_x</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>除非你将分支推送到远端仓库，不然该分支就是 不为他人所见的：</p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git push origin <branch></span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n<h2 id=\"多人协同完整流程\"><a href=\"#多人协同完整流程\" class=\"headerlink\" title=\"多人协同完整流程\"></a>多人协同完整流程</h2><ul>\n<li>GitLab 新建远程分支 <code>develop</code></li>\n</ul>\n<blockquote>\n<p>这一步的意义在于防止直接操作 <code>master</code> 分支造成无法挽回的后果</p>\n</blockquote>\n<ul>\n<li><p>下载新的远程分支</p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git fetch origin</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>值得注意的是，在 <code>fetch</code> 操作下载好新的远程分支之后，你仍然无法在本地编辑该远程仓库中的分支。换句话说，在本例中，你不会有一个新的 <code>develop</code> 分支，有的只是一个你无法移动的 <code>origin/develop</code> 指针</p>\n</blockquote>\n</li>\n<li><p>操作下载的远程分支</p>\n<ul>\n<li><p>如果要把该远程分支的内容合并到当前分支，可以运行</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git merge origin/develop</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>如果想要一份自己的 serverfix 来开发，可以在远程分支的基础上分化出一个新的分支来</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git checkout -b develop origin/develop</span><br></pre></td></tr></table></figure>\n<p>这会切换到新建的 <code>develop</code> 本地分支，其内容同远程分支 <code>origin/develop</code> 一致，这样你就可以在里面继续开发了。</p>\n</li>\n</ul>\n<blockquote>\n<p>推荐</p>\n</blockquote>\n</li>\n<li><p>新建本地用户分支并开发</p>\n <figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git checkout -b holmeyoung</span><br></pre></td></tr></table></figure>\n<p> 在 <code>holmeyoung</code> 分支开发完毕后</p>\n<ul>\n<li><p>提交到缓存区</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git add --all</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>提交到 HEAD</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git commit -m <span class=\"string\">\"holmeyoung change\"</span></span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n</li>\n<li><p>合并改动到 <code>develop</code> 分支</p>\n<ul>\n<li><p>跳到 <code>develop</code> 分支</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git checkout develop</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>将 <code>holmeyoung</code> 分支的改动 merge 到 <code>develop</code> 分支</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git merge holmeyoung</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>推送本地 <code>develop</code> 分支到远程 <code>develop</code> 分支</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git push origin develop:develop</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n</li>\n<li><p>合并改动到 <code>master</code> 分支</p>\n<ul>\n<li><p>跳到 <code>master</code> 分支</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git checkout master</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>将 <code>develop</code> 分支的改动 merge 到 <code>master</code> 分支</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git merge develop</span><br></pre></td></tr></table></figure>\n</li>\n<li><p>推送本地 <code>master</code> 分支到远程 <code>master</code> 分支</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\">git push origin master:master</span><br></pre></td></tr></table></figure>\n</li>\n</ul>\n</li>\n</ul>\n<h2 id=\"REF\"><a href=\"#REF\" class=\"headerlink\" title=\"REF\"></a><code>REF</code></h2><blockquote>\n<p><a href=\"http://www.bootcss.com/p/git-guide/\" target=\"_blank\" rel=\"noopener\">git - 简易指南</a></p>\n<p><a href=\"http://www.ruanyifeng.com/blog/2014/06/git_remote.html\" target=\"_blank\" rel=\"noopener\">Git远程操作详解</a></p>\n<p><a href=\"https://git-scm.com/book/zh/v2\" target=\"_blank\" rel=\"noopener\">Git Book</a></p>\n</blockquote>\n","categories":["技术总结"],"tags":["git"]},{"title":"anaconda 安装记录","url":"https://www.holmeyoung.com/blog/anaconda-linux/","content":"<h3 id=\"Anaconda-是什么？\"><a href=\"#Anaconda-是什么？\" class=\"headerlink\" title=\"Anaconda 是什么？\"></a>Anaconda 是什么？</h3><p>Anaconda 是一个可用于科学计算的 Python 发行版，支持 Linux、Mac、Windows系统，内置了常用的科学计算包。它解决了官方 Python 的两大痛点。</p>\n<ul>\n<li>第一：提供了包管理功能，Windows 平台安装第三方包经常失败的场景得以解决</li>\n<li>第二：提供环境管理的功能，功能类似 Virtualenv，解决了多版本Python并存、切换的问题</li>\n</ul>\n<h3 id=\"下载-Anaconda\"><a href=\"#下载-Anaconda\" class=\"headerlink\" title=\"下载 Anaconda\"></a>下载 Anaconda</h3><p>直接在<a href=\"https://www.anaconda.com/download/#macos\" target=\"_blank\" rel=\"noopener\">官网下载</a>安装包，我是下载的 <code>2.7</code> 版本。下载后直接安装并选择默认配置就好。</p>\n<p>安装过程会提示安装 <code>visual studio code</code> 选择安装，如果说为什么的话</p>\n<blockquote>\n<p>比sublime开源、比atom更快、比webstorm更轻</p>\n</blockquote>\n<h3 id=\"环境变量配置\"><a href=\"#环境变量配置\" class=\"headerlink\" title=\"环境变量配置\"></a>环境变量配置</h3><h4 id=\"如何配置\"><a href=\"#如何配置\" class=\"headerlink\" title=\"如何配置\"></a>如何配置</h4><p>刚下载下来在 <code>iterm2</code> 中输入 <code>conda</code> 会显示 <code>zsh command not found</code></p>\n<p>因为是我用的是 <code>iterm2</code> 且配置了 <code>oh my zsh</code> ，错误显示的又是 <code>zsh</code> 所以我们需要在 <code>~/.zsh</code> 即 <code>oh my zsh</code> 的配置文件中添加 <strong>anaconda</strong> 的安装路径。</p>\n<p>即添加</p>\n<figure class=\"highlight sh\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"built_in\">export</span> PATH=<span class=\"string\">\"/anaconda2/bin:<span class=\"variable\">$PATH</span>\"</span></span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>需要注意自己用户目录下的是 <code>anaconda2</code> 还是 <code>anaconda</code> </p>\n</blockquote>\n<h4 id=\"检查配置\"><a href=\"#检查配置\" class=\"headerlink\" title=\"检查配置\"></a>检查配置</h4><p>输入 <code>which conda</code> ,返回具体的路径即为成功！</p>\n<h3 id=\"conda-命令介绍\"><a href=\"#conda-命令介绍\" class=\"headerlink\" title=\"conda 命令介绍\"></a>conda 命令介绍</h3><h4 id=\"关于新建Python\"><a href=\"#关于新建Python\" class=\"headerlink\" title=\"关于新建Python\"></a>关于新建Python</h4><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 查看帮助</span></span><br><span class=\"line\">conda -h</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 基于python3.6版本创建一个名字为python36的环境</span></span><br><span class=\"line\">conda create --name python36 python=3.6</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 激活此环境 linux/mac</span></span><br><span class=\"line\">source activate python36 </span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 再来检查python版本，显示是 3.6</span></span><br><span class=\"line\">python -V</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 退出当前环境到系统根目录</span></span><br><span class=\"line\">source deactivate</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 删除该环境</span></span><br><span class=\"line\">conda remove -n python36 --all</span><br><span class=\"line\"></span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 查看所有安装的环境</span></span><br><span class=\"line\">conda info -e</span><br><span class=\"line\"></span><br><span class=\"line\">➜ ~ conda info -e</span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> conda environments:</span></span><br><span class=\"line\"><span class=\"meta\">#</span></span><br><span class=\"line\">base * /Users/holmeyoung/anaconda2</span><br><span class=\"line\">python36 /Users/holmeyoung/anaconda2/envs/python36</span><br></pre></td></tr></table></figure>\n<h4 id=\"关于conda-的包管理\"><a href=\"#关于conda-的包管理\" class=\"headerlink\" title=\"关于conda 的包管理\"></a>关于conda 的包管理</h4><p>和 <code>pip</code> 是一样的，当然你选择 <code>pip</code> 来安装包也是没问题的</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 安装 matplotlib </span></span><br><span class=\"line\">conda install matplotlib</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 查看已安装的包</span></span><br><span class=\"line\">conda list</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 包更新</span></span><br><span class=\"line\">conda update matplotlib</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 删除包</span></span><br><span class=\"line\">conda remove matplotlib</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>在 conda 中 <strong>anything is a package</strong>。conda 本身可以看作是一个包，python 环境可以看作是一个包，anaconda 也可以看作是一个包，因此除了普通的第三方包支持更新之外……</p>\n</blockquote>\n<h4 id=\"conda的自我更新\"><a href=\"#conda的自我更新\" class=\"headerlink\" title=\"conda的自我更新\"></a>conda的自我更新</h4><figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 更新conda本身</span></span><br><span class=\"line\">conda update conda</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 更新anaconda 应用</span></span><br><span class=\"line\">conda update anaconda</span><br><span class=\"line\"></span><br><span class=\"line\"><span class=\"meta\">#</span><span class=\"bash\"> 更新python，假设当前python环境是3.6.1，而最新版本是3.6.2，那么就会升级到3.6.2</span></span><br><span class=\"line\">conda update python</span><br></pre></td></tr></table></figure>\n<h4 id=\"anaconda卸载\"><a href=\"#anaconda卸载\" class=\"headerlink\" title=\"anaconda卸载\"></a>anaconda卸载</h4><p>由于Anaconda的安装文件都包含在一个目录中，所以直接将该目录删除即可</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">rm -rf anaconda2</span><br></pre></td></tr></table></figure>\n<blockquote>\n<p>注意清理 <code>.zshrc</code> 以及 <code>~/.bash_profile</code> 中的Anaconda路径</p>\n</blockquote>\n<h3 id=\"部分参考\"><a href=\"#部分参考\" class=\"headerlink\" title=\"部分参考\"></a>部分参考</h3><p><a href=\"https://foofish.net/anaconda-install.html\" target=\"_blank\" rel=\"noopener\">Anaconda 入门安装教程</a></p>\n<p><a href=\"http://www.cnblogs.com/caiyutong/p/7002764.html\" target=\"_blank\" rel=\"noopener\">Mac下安装Anaconda和环境变量设置，以及升级和卸载</a></p>\n","categories":["安装记录"],"tags":["python","linux"]},{"title":"实现 ssh 免密码登录服务器","url":"https://www.holmeyoung.com/blog/ssh-no-password/","content":"<h3 id=\"修改-ssh-config权限\"><a href=\"#修改-ssh-config权限\" class=\"headerlink\" title=\"修改 ~/.ssh/config权限\"></a>修改 <code>~/.ssh/config</code>权限</h3><p><code>~/.ssh/config</code>是 ssh 自己的配置文件，在修改文件之前一定要修改文件的权限以及其所属于的用户，否则会出现问题。</p>\n<p>执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">chmod 600 ~/.ssh/config</span><br></pre></td></tr></table></figure>\n<p>否则可能出现</p>\n<blockquote>\n<p><strong>Bad owner or permissions on /Users/mac/.ssh/config</strong></p>\n</blockquote>\n<p>执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">chown $USER ~/.ssh/config</span><br></pre></td></tr></table></figure>\n<p>否则可能出现</p>\n<blockquote>\n<p><strong>ssh: Could not resolve hostname centos: nodename nor servname provided, or not known</strong></p>\n</blockquote>\n<p>确定 <strong>USER</strong> 的方法是执行 </p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">whoami</span><br></pre></td></tr></table></figure>\n<p> 如我的结果是 <code>mac</code> 即执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">chown mac ~/.ssh/config</span><br></pre></td></tr></table></figure>\n<h3 id=\"修改-ssh-config\"><a href=\"#修改-ssh-config\" class=\"headerlink\" title=\"修改 ~/.ssh/config\"></a>修改 <code>~/.ssh/config</code></h3><p>执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">vim ~/.ssh/config</span><br></pre></td></tr></table></figure>\n<p>将下面内容添加进去</p>\n<figure class=\"highlight powershell\"><table><tr><td class=\"code\"><pre><span class=\"line\">Host 起个名字</span><br><span class=\"line\"> HostName 你的服务器IP</span><br><span class=\"line\"> Port ssh端口</span><br><span class=\"line\"> User ssh用户</span><br><span class=\"line\"> IdentityFile 私钥文件路径(~/.ssh/id_rsa)</span><br></pre></td></tr></table></figure>\n<p>保存退出</p>\n<h3 id=\"设置免密码登录\"><a href=\"#设置免密码登录\" class=\"headerlink\" title=\"设置免密码登录\"></a>设置免密码登录</h3><h4 id=\"本地生成公钥和私钥文件\"><a href=\"#本地生成公钥和私钥文件\" class=\"headerlink\" title=\"本地生成公钥和私钥文件\"></a>本地生成公钥和私钥文件</h4><p>在 Mac OSX 终端（或 iTerm 2 等）中进入 <code>~/.ssh</code> 目录，输入</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">ssh-keygen -t rsa</span><br></pre></td></tr></table></figure>\n<p>会问一些问题，一路回车就好啦。这样就会在 <code>~/.ssh</code> 目录中生成 <code>id_rsa</code>（私钥） 和 <code>id_rsa.pub</code>（公钥）文件</p>\n<h4 id=\"服务器端配置\"><a href=\"#服务器端配置\" class=\"headerlink\" title=\"服务器端配置\"></a>服务器端配置</h4><p>进入服务器端 <code>~/.ssh</code> 目录，执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">vim authorized_keys</span><br></pre></td></tr></table></figure>\n<p>没有的话新建就好了</p>\n<p>打开Mac端 <code>~/.ssh</code> 目录中生成的 <code>id_rsa.pub</code> 文件，将其中的内容复制到服务器端的 <code>authorized_keys</code> 文件中</p>\n<p>保存退出即可</p>\n<blockquote>\n<p>要是有其他用户已经新建了这个文件并且写入了内容，只需要在文件最后追加即可</p>\n</blockquote>\n<h4 id=\"修改文件权限\"><a href=\"#修改文件权限\" class=\"headerlink\" title=\"修改文件权限\"></a>修改文件权限</h4><p>确保下文件权限正常，在服务器端执行</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">chmod 700 ~/.ssh/</span><br><span class=\"line\">chmod 600 ~/.ssh/authorized_keys</span><br></pre></td></tr></table></figure>\n<p>如果无法连接，请在服务器的 <code>/etc/ssh/sshd_config</code> 文件中查看 <code>PubkeyAuthentication</code> 的值是否为 <strong>yes</strong> ，如果不是，请修改为 <strong>yes</strong> ，并使用以下命令重启 ssh 服务</p>\n<figure class=\"highlight powershell\"><table><tr><td class=\"code\"><pre><span class=\"line\">/etc/init.d/ssh restart</span><br></pre></td></tr></table></figure>\n<p>如果需要禁止密码登录，同样可以编辑 <code>/etc/ssh/sshd_config</code> 文件中 <code>PasswordAuthentication</code> 的值为 <strong>no</strong> 并重启 ssh 服务</p>\n<h3 id=\"最终效果\"><a href=\"#最终效果\" class=\"headerlink\" title=\"最终效果\"></a>最终效果</h3><p>例如你在 <code>~/.ssh/config</code> 中写入的是</p>\n<figure class=\"highlight powershell\"><table><tr><td class=\"code\"><pre><span class=\"line\">Host centos</span><br><span class=\"line\"> HostName <span class=\"number\">8.8</span>.<span class=\"number\">8.8</span></span><br><span class=\"line\"> Port <span class=\"number\">22</span></span><br><span class=\"line\"> User root</span><br><span class=\"line\"> IdentityFile ~/.ssh/id_rsa</span><br></pre></td></tr></table></figure>\n<p>那么在终端或者 iterm 中输入</p>\n<figure class=\"highlight shell\"><table><tr><td class=\"code\"><pre><span class=\"line\">ssh centos</span><br></pre></td></tr></table></figure>\n<p>即可直接登录服务器啦</p>\n","categories":["技术总结"],"tags":["linux","ssh"]},{"title":"about","url":"https://www.holmeyoung.com/about/index.html","content":"","categories":[],"tags":[]},{"title":"category","url":"https://www.holmeyoung.com/category/index.html","content":"","categories":[],"tags":[]},{"title":"link","url":"https://www.holmeyoung.com/link/index.html","content":"","categories":[],"tags":[]},{"title":"project","url":"https://www.holmeyoung.com/project/index.html","content":"","categories":[],"tags":[]},{"title":"search","url":"https://www.holmeyoung.com/search/index.html","content":"","categories":[],"tags":[]},{"title":"tag","url":"https://www.holmeyoung.com/tag/index.html","content":"","categories":[],"tags":[]}]