The latest articles on DEV Community by TianZhen (@dawncold). https://dev.to/dawncold https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F32749%2F6e16b65b-af30-481f-9144-d41a1cb450aa.jpeg https://dev.to/dawncold en TianZhen Sat, 09 Sep 2017 02:59:20 +0000 https://dev.to/dawncold/new-to-devto-how-it-works-so-fast https://dev.to/dawncold/new-to-devto-how-it-works-so-fast <p><a href="//dev.to">dev.to</a> is so fast for me, I'm wondering how it works so fast, after I registered and write a post, I know some of it.</p> <p>The posts are written in <code>Markdown</code> and generated to a static file, just like some static-blog-generator, they have same syntax: some mate info at the beginning of post.</p> <p>But I love that, hope someday I could export all my posts and build a personal blog.</p> TianZhen Sat, 02 Sep 2017 08:10:10 +0000 https://dev.to/dawncold/job-queue-rollback https://dev.to/dawncold/job-queue-rollback <h2> 背景 </h2> <p>我们的异步队列是<a href="https://github.com/closeio/tasktiger">Tasktiger</a>，一个背后有商业公司支持的开源项目。</p> <p>之前是<a href="https://github.com/binarydud/pyres">Pyres</a>，一个<a href="https://github.com/resque/resque">Resque</a>çš„Pythonå®žçŽ°ï¼Œå› ä¸º<a href="https://github.com">Github</a>在使用<a href="https://github.com/resque/resque">Resque</a>，<a href="https://www.instagram.com/">Instagram</a>在使用<a href="https://github.com/binarydud/pyres">Pyres</a>，我们应该可以用一段时间。</p> <p>后来<a href="https://github.com/binarydud/pyres">Pyres</a>长久不再维护，积累了一些bug以及性能问题让我们不得不选择更换一个异步队列系统，期初考虑过<a href="http://www.celeryproject.org/">Celery</a>，那是一个很成熟很强大的队列系统，以至于我很难一下子搞清它。</p> <h2> 问题 </h2> <p>在使用<code>Tasktiger</code>后有时会遇到已经处理完毕的task又重新被enqueue的情况，如果所有task都是幂等的（Idempotent），那这也不是问题，第二次执行taskä¼šå› ä¸ºä¹‹å‰å¤„ç†è¿‡è€Œä¸è‡³äºŽå†åšä¸€æ¬¡ã€‚ç„¶è€Œå¹¶ä¸æ˜¯æ‰€æœ‰taskéƒ½å®Œæˆäº†å¹‚ç­‰æ”¹é€ ï¼Œæ‰€ä»¥è¿™å¯¹äºŽæˆ‘ä»¬æ¥è¯´æ˜¯ä¸ªä¸¥é‡é—®é¢˜ï¼Œä½†å½“æ—¶å¹¶ä¸èƒ½ç¨³å®šé‡çŽ°ã€‚</p> <h2> 探索 </h2> <p>怀疑过许多地方，后来发现在queue redis上执行<code>BGREWRITEAOF</code>就能避免这个问题，于是把原先一个月执行一次的<code>AOF</code>é‡å†™æ“ä½œå˜æˆäº†æ¯å¤©æ‰§è¡Œä¸€æ¬¡ï¼Œå¤§æ¦‚åœ¨å¤‡ä»½æ•°æ®ä¹‹å‰ï¼Œå› ä¸ºå¤‡ä»½éœ€è¦åœæœºï¼Œå¦‚æžœä¸åœ¨è¿™ä¹‹å‰åšï¼Œå¯èƒ½å¤‡ä»½å¯¼è‡´é‡å¯æœåŠ¡å°±ä¼šè®©å¾ˆå¤šå·²ç»å¤„ç†è¿‡çš„task重复处理，例如可能会给会员发多条短信和微信提醒。</p> <p>å°±åœ¨æ˜¨å¤©ï¼Œå› ä¸ºé‡å¯queue redis导致很多task重复处理，这个问题终究没有被解决，于是花了大约2å¤©æ—¶é—´ï¼Œæ‰¾åˆ°äº†å¯¼è‡´é—®é¢˜çš„æ ¹æºã€‚</p> <p>之前猜测是不是<code>Tasktiger</code>哪里没写好导致，毕竟这个项目在Github上只有不到500个star，而redisæœ‰å‡ ä¸‡ï¼Œå¹¿å¤§ç”¨æˆ·åŸºç¡€æ›´å®¹æ˜“å‘çŽ°å¥‡æ€ªbug，而在<code>Tasktiger</code>上遇到某个bugï¼Œä¹Ÿè®¸ä½ å°±æ˜¯ä¸–ç•Œä¸Šç¬¬ä¸€ä¸ªé‡åˆ°çš„ã€‚åŸºäºŽè¿™ä¸ªå‡è®¾æˆ‘çœ‹äº†å¾ˆå¤šæºä»£ç ï¼Œæ²¡æœ‰æ‰¾åˆ°å¯é è¯æ®ï¼Œè€Œä¸”åŸºäºŽä¸€ä¸ªç®€å•äº‹å®žï¼šä¸€ä¸ªtask从最开始不存在，到出现在Tasktigerä¸­ï¼Œåˆ°å¤„ç†å®Œè¢«åˆ é™¤ï¼Œå¦‚æžœæ˜¯<code>Tasktiger</code>çš„é—®é¢˜ï¼Œé‚£å®ƒä»Žå“ªé‡Œå¾—åˆ°æ•°æ®æž„é€ å‡ºä¸€ä¸ªtask呢？难道是redis的问题吗？</p> <p>queue redis一般都会启用<code>AOF</code>，所有修改数据的命令和值被记录到<code>AOF</code>中，等redis重启时，会执行<code>AOF</code>æ–‡ä»¶ä¸­çš„å‘½ä»¤ï¼Œè¿™æ ·rediså°±æœ‰äº†å’Œé‡å¯å‰ä¸€æ¨¡ä¸€æ ·çš„æ•°æ®ã€‚åœ¨<code>AOF</code>中记录着某个task从出现到消失的全过程，难道是redis没完成loadå°±æŽ¥å—æœåŠ¡äº†ï¼Ÿä»Žæ—¥å¿—ä¸Šçœ‹å¹¶ä¸æ˜¯è¿™æ ·ï¼Œåœ¨æ²¡load完执行操作会得到异常。</p> <p>看了下生产环境的queue状态，基本没有正在处理的task，停掉queue redis，下载<code>AOF</code>文件，在本地让redis启动并load这份<code>AOF</code>，不启动<code>Tasktiger</code>çš„worker，只静静地看queue的状态，发现load完后有很多task被重新放回了queue，此时如果有worker，一定会重复执行。</p> <p>到这里难道能确定是redis的问题吗？redisçš„<code>AOF</code>并不能让rediså›žåˆ°ä¹‹å‰çš„çŠ¶æ€ï¼Ÿå¦‚æžœæ˜¯è¿™æ ·çš„è¯redis怎么会被广泛应用？</p> <p>于是找了一个task，从<code>AOF</code>中看看<code>Tasktiger</code>是如何处理它的。</p> <p>处理过程没发现异常，只是<code>Tasktiger</code>用了很多<code>lua</code>脚本操作数据，例如有个脚本是：当一个keyä¸åœ¨ç»™å®šçš„æŸå‡ ä¸ªzsetä¸­æ—¶åˆ é™¤è¿™ä¸ªkeyã€‚è¿™ä¸ªæ“ä½œä¸å¤æ‚ï¼Œä½†ä½ æ— æ³•åªç”¨rediså®Œæˆï¼Œå› ä¸ºredisåªæ˜¯ä¸€ä¸ªæ•°æ®å­˜å‚¨ï¼Œæ— æ³•å¤„ç†å¤æ‚é€»è¾‘ï¼Œæ‰€ä»¥åªèƒ½ä½¿ç”¨<code>lua</code>完成，当然很多语都可以完成这个逻辑，只不过redis支持执行<code>lua</code>脚本，算作是对redis操作的补充。</p> <p>å› ä¸ºæœ‰äº†ä¸€ä¸ªå…·ä½“task，查看和它相关的所有操作后，发现除非是在某一步<code>lua</code>脚本未执行时才会导致这个task被放在active queue中而不是被移除。</p> <p>äºŽæ˜¯æˆ‘æž„é€ äº†ä¸€ä»½<code>AOF</code>ï¼Œæœ€å¼€å§‹çš„éƒ¨åˆ†æ˜¯ä¸€å †<code>SCRIPT LOAD</code>指令，把<code>Tasktiger</code>需要的脚本都load到redisä¸­ï¼Œè¿™æ ·åŽç»­æ“ä½œæ‰å¯ä»¥ä½¿ç”¨<code>EVALSHA</code>ï¼Œè¿™ä»½æž„é€ å¥½çš„<code>AOF</code>也通过了<code>redis-check-aof</code>检查，被load后再次查看queue状态，竟然正常了，和停redis之前一致！看来问题出在执行这些脚本的地方。</p> <p>在redis中执行<code>lua</code>脚本有两种做法，第一种是使用<code>EVAL</code>ï¼Œè¿™ç§åšæ³•æ˜¯æŠŠè„šæœ¬å†…å®¹ä¸€å¹¶è´´ä¸Šï¼Œæ¯æ¬¡æ‰§è¡Œéƒ½éœ€è¦è¿™æ ·ï¼Œå¦‚æžœæ€»æ˜¯é‡å¤æ‰§è¡ŒæŸä¸ªç‰¹å®šè„šæœ¬å°±ä¼šå› ä¸ºä¼ è¾“äº†å¤§é‡è„šæœ¬å†…å®¹è€Œä½¿å¾—æ•ˆçŽ‡ä¸å¤Ÿé«˜ï¼ŒäºŽæ˜¯æœ‰äº†ç¬¬äºŒç§æ‰§è¡Œæ–¹æ³•––<code>EVALSHA</code>，使用这种需要脚本的<code>SHA1</code>，当执行<code>SCRIPT LOAD</code>后，redis返回的内容就是脚本的<code>SHA1</code>，load过的脚本在手工flush前或者redis重启前都是有效的。<code>SCRIPT LOAD</code>在<code>BGREWRITEAOF</code>åŽå¹¶ä¸ä¼šå­˜åœ¨ï¼Œè™½ç„¶ä½ è¿˜æ˜¯å¯ä»¥é€šè¿‡<code>EVALSHA</code>调用脚本。</p> <p>那么这两种执行方法对于<code>AOF</code>文件记录内容有差别吗？没有，执行<code>EVALSHA</code>也会被记录成<code>EVAL</code>ï¼Œå¦‚æžœä½ åœ¨<code>AOF</code>中发现有<code>EVALSHA</code>那就奇怪了。在生产环境下载的<code>AOF</code>中，有不少<code>EVALSHA</code>，为何没有被记录成<code>EVAL</code>ï¼Ÿå› ä¸ºè¿™äº›<code>EVALSHA</code>并不是redis执行的，而是<code>Tasktiger</code>çš„<code>pipeline</code></p> <p><strong>å¯¼è‡´å‡ºçŽ°å›žæ»šé—®é¢˜çš„åŽŸå› æ˜¯ï¼šBGREWRITEAOF后，由于SCRIPT LOAD被去掉，redis重启后如果有使用EVALSHAè°ƒç”¨è„šæœ¬çš„åœ°æ–¹ä¼šå‡ºé”™ï¼Œå› ä¸ºé‚£ä¸ªè„šæœ¬å¹¶æ²¡æœ‰è¢«redis load</strong></p> <p>不是说<code>EVALSHA</code>会被翻译成<code>EVAL</code>放在<code>AOF</code>中吗？是的，不过<code>Tasktiger</code>自己实现了一个redis <code>pipeline</code>，这也是个<code>lua</code>脚本，在这个<code>lua</code>脚本中是需要调用其他<code>lua</code>脚本的。调用方法包含两种，大多数都是<code>EVALSHA</code>， <strong>由于redis禁止在lua中使用redis.call执行EVALSHA调用另一个脚本（并不是所有redis命令都可以在redis.call使用）</strong> ，Tasktiger使用了其他方法来调用，算是一种hack。</p> <p>由于redisçš„é™åˆ¶åŠ ä¸Š<code>Tasktiger</code>çš„hack，导致 <strong>在BGREWRITEAOF后，如果有pipeline脚本的调用发生，下次重启redis会使得之前做过的task被重新放回queueï¼Œå› ä¸ºç§»é™¤æ“ä½œçš„è„šæœ¬è¿˜æ²¡æœ‰load，所以移除失败了，看起来就像是task被重新放回queue</strong></p> <p><strong>目前我们针对这个问题的解决方法是：勤快点BGREWRITEAOF，每小时一次，重启redis之前再做一次</strong></p> redis tasktiger TianZhen Sat, 02 Sep 2017 06:22:19 +0000 https://dev.to/dawncold/worst-day https://dev.to/dawncold/worst-day <h2> 21:40 - 服务不可用 </h2> <p>监控宝提示Homepageæ— æ³•è®¿é—®ï¼Œä»¥ä¸ºæœ‰äººåœ¨éƒ¨ç½²ï¼ŒåŽæ¥å…¶ä»–æœåŠ¡ä¹Ÿä¸å¯ç”¨ï¼Œç™»å½•åˆ°æœåŠ¡å™¨ä¸Šçœ‹å‘çŽ°<code>top</code>都不能执行了，报过<code>Bus error</code>å’Œ<code>Segment fault</code>等，看<code>/var/log/syslog</code>发现是<code>RAID</code>卡出错，从<code>iDRAC</code>上看没有错误，<code>RAID</code>å¡å’Œç”µæ± çŠ¶æ€æ­£å¸¸ã€‚</p> <p>ç›®å‰å·²ç»æ— æ³•æä¾›æœåŠ¡äº†ï¼Œé‡å¯è¯•ä¸€ä¸‹ã€‚</p> <p><code>RAID</code>å¡æ— æ³•è¢«è¯†åˆ«åˆ°ï¼Œç³»ç»Ÿè¿›ä¸åŽ»ã€‚</p> <p>强制刷一下<code>RAID</code>å¡å›ºä»¶ï¼Œå¤±è´¥ï¼Œéƒ½æ— æ³•è¯†åˆ«ï¼Œä¹Ÿæ²¡æ³•åˆ·å›ºä»¶äº†ã€‚</p> <p>å…³æœºï¼Œç­‰ç¬¬äºŒå¤©åŽ»æœºæˆ¿æŠŠç£ç›˜æ¢åˆ°å¦ä¸€å°æœºå™¨ä¸Šï¼Œå¦ä¸€å°æœºå™¨æœ‰åž‹å·ä¸€æ ·çš„<code>RAID</code>卡。</p> <h2> 7:00 - 能进系统了 </h2> <p>ç¬¬äºŒå¤©æ—©æ™¨å¼€æœºåŽå‘çŽ°å¯ä»¥è¿›ç³»ç»Ÿï¼Œä½†æ•°æ®åº“å¯åŠ¨ä¸äº†ï¼Œå¯èƒ½æ•°æ®åº“çš„æŸäº›æ–‡ä»¶è¢«ç ´åã€‚</p> <p>å†æ¬¡é‡å¯æœºå™¨ï¼Œç³»ç»Ÿæ— æ³•å¯åŠ¨ï¼Œéœ€è¦<code>fsck</code></p> <h2> 8:10 - 进入机房 </h2> <p>去机房把服务器电源拔掉，长按开机键让电量耗光，如果这种reset还是没能解决问题，就只能更换到另一台服务器了。</p> <p>hardware reset不管用，把磁盘放入另一台服务器。</p> <p>启动时提醒需要import raid配置，import进入系统提示需要<code>fsck</code></p> <p>手工执行<code>fsck</code>，出现很多错误，一一确认后让<code>fsck</code>修复，可以进入系统。</p> <p>备份系统到别的服务器</p> <h2> 14:30 - 修复数据库 </h2> <p>数据库（PostgreSQL 9.6）和一些静态文件有损坏</p> <p>备份了数据库的<code>data</code>目录后，开始试着恢复数据库</p> <p>reset了xlog，用<code>dd</code>创建了一个全<code>0</code>çš„clog，数据库可以启动，但有些奇怪的数据，<code>dump</code>出所有数据重建<code>database</code>再<code>restore</code>ï¼Œå› ä¸ºæˆ‘ä»¬æ¯”è¾ƒåœ¨æ„database的建模，很多有关系的表之间都有外键约束，<code>restore</code>的时候能帮我们发现很多不一致的数据，只好fix-dump-restore测试，一直到所有数据都正常后，<code>vacuum</code>， rebuild index</p> <h2> 21:00 - 结束 </h2> <p>真是worst dayï¼ŒçŽ°åœ¨çœ‹æ¥æ•°æ®åº“æ–¹é¢æœ‰æ›´å¥½çš„ä¿®å¤æ–¹æ³•ï¼Œèµ·ç å¯ä»¥åšåˆ°åªä¸¢å¤±æœ€è¿‘çš„ä¸€éƒ¨åˆ†æ•°æ®ï¼Œç„¶è€Œæˆ‘ä»¬æ²¡æœ‰realtime的数据库备份也是个大问题，准备使用<a href="http://www.pgbarman.org/">Barman</a>来解决这些问题，另外服务器硬件也有一些东西需要重新考虑</p> postgres hardware TianZhen Thu, 24 Aug 2017 13:09:36 +0000 https://dev.to/dawncold/unlock-hg8347r https://dev.to/dawncold/unlock-hg8347r <p>联通宽带移机一定要带着旧光猫，不管旧光猫是好是坏是死是活。* 3</p> <p>æ–°å®¶å®Œæˆå®½å¸¦ç§»æœºï¼Œä¸æ”¶è´¹ï¼Œä½†å¦‚æžœä½ æ²¡æœ‰æ—§å…‰çŒ«çš„è¯ï¼Œå¾—è‡ªå·±è§£å†³ï¼Œæ¯”å¦‚æ·˜å®ä¸€ä¸ªå¤§çº¦50块。</p> <p>新光猫是HG8347Rï¼Œæ®è¯´ç¡¬ä»¶é…ç½®ä¸é”™ï¼Œä½†æˆ‘å…³å¿ƒçš„æ˜¯èƒ½å¦ç»§ç»­ä½¿ç”¨è‡ªå·±çš„è·¯ç”±å™¨ï¼Œé»˜è®¤æƒ…å†µä¸‹ä¸èƒ½ï¼Œå› ä¸ºè¿™æ¬¾å…‰çŒ«é›†æˆäº†æ— çº¿è·¯ç”±åŠŸèƒ½ã€‚å¯¹äºŽå¤§å¤šæ•°ç”¨æˆ·æ¥è®²æ˜¯ä¸ªå¥½äº‹å„¿ï¼Œä½†å¯¹äºŽå…¶ä»–ç”¨æˆ·ï¼Œä»–ä»¬å°±ä¸è¿‡å¤šè€ƒè™‘äº†ï¼Œå¥½åœ¨å…‰çŒ«æ²¡é”æ­»ï¼Œå¯ä»¥è‡ªå·±è§£é”ã€‚</p> <p>éœ€è¦æœ‰å‡ æ¬¾è½¯ä»¶ï¼šåŽä¸ºONT组播维护工具、配置文件解密工具、tftp server，可从<a href="https://pan.baidu.com/s/1pLfYcJ5">这里</a>ä¸‹è½½åˆ°ã€‚ç»´æŠ¤å·¥å…·ä½œç”¨æ˜¯å¼€å¯å…‰çŒ«ç»´æŠ¤æ¨¡å¼ï¼Œè¿™æ ·æ‰å¯ä»¥telnet进去，拿到配置文件、恢复出厂设置等。</p> <h2> 开启维护模式 </h2> <ol> <li>电脑网线连接光猫LAN1口</li> <li>拔掉光猫光纤，此时光猫PON位置亮红灯</li> <li>打开ONT组播工具，选择连接光猫的网卡，点启动</li> <li>等到光猫所有指示灯全亮后，关闭光猫电源，此时维护模式已开启。如果长时间没进入维护模式（比如超过了2分钟）可以尝试关闭光猫电源后再打开</li> </ol> <h2> 获取配置文件 </h2> <ol> <li>启动已开启维护模式的光猫</li> <li>电脑可以ping通光猫ip后，打开命令行</li> <li> <code>telnet 192.168.1.1</code>，用户名是：rootï¼Œå¯†ç æ˜¯ï¼šadmin</li> <li>登录后在电脑开启<code>tftp server</code> </li> <li>在命令行中输入：<code>backup cfg by tftp svrip 192.168.1.2 remotefile hw_ctree.xml</code>,其中<code>192.168.1.2</code>是电脑IP，命令完成后可在<code>tftp</code>设置的目录中找到<code>hw_ctree.xml</code>文件</li> <li>è¿™æ˜¯ä¸€ä¸ªåŠ å¯†åŽçš„gzip文件，此时需要用到第二个软件：配置文件解密工具</li> <li>解密后得到的文件是gzip文件，用gzip工具解压可得到真正<code>hw_ctree.xml</code> </li> <li>在配置文件中找到<code>UserLevel="1"</code>，将其修改为<code>0</code>，表示特权用户</li> </ol> <h2> 恢复出厂设置 </h2> <p>回到telnet开始恢复出厂设置<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code>su shell restorehwmode.sh exit reset </code></pre> </div> <p>如果设备没重启则可以手动关掉电源再打开</p> <h2> 导入配置 </h2> <ol> <li>设备重启后修改电脑IP为<code>192.168.100.2</code>，网关是<code>192.168.100.1</code> </li> <li>访问<code>192.168.100.1</code>，使用默认用户名：<code>telecomadmin</code>ï¼Œå¯†ç ï¼š<code>admintelecom</code>登录，导入刚刚修改了配置文件，设备重启后即可自动获取IP地址</li> </ol> <h2> 设置WAN桥接 </h2> <ol> <li>访问<code>192.168.1.1</code>ï¼Œå› ä¸ºé…ç½®æ–‡ä»¶åªä¿®æ”¹äº†ç”¨æˆ·æƒé™æ²¡æœ‰ä¿®æ”¹è´¦å·å¯†ç ï¼Œæ‰€ä»¥å¯ç”¨å…‰çŒ«èƒŒåŽçš„ç”¨æˆ·åå¯†ç ç™»å½•</li> <li>在WAN设置中找到<code>2_INTERNET_B_VID_3961</code>，点击一下可以展开编辑，修改<code>WAN类型</code>为<code>WAN桥接</code>即可使用自己路由器，路由器WAN口接光猫LAN1口，使用路由器认证。使用SPEEDTEST测试速度在100Mbps，没掉速。</li> </ol> hg8347r chinaunicom TianZhen Thu, 24 Aug 2017 06:35:36 +0000 https://dev.to/dawncold/how-to-understand-program-to-an-interface-not-an-implementation https://dev.to/dawncold/how-to-understand-program-to-an-interface-not-an-implementation <p>Yesterday my colleagues and I talk about JS <code>for loop</code> and <code>const</code>. I found this <a href="https://medium.com/javascript-scene/master-the-javascript-interview-what-is-a-closure-b2f0d2152b36">post</a>, it refers a quote "Program to an interface not an implementation", I read this several years ago, and have no feeling about this, just think it's a cool quote or rule.</p> <p>Now, after several years working experience, I got some understanding about that.</p> <p>For instance I use a simple JS example,<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight javascript"><code><span class="kd">var</span> <span class="nx">counter</span> <span class="o">=</span> <span class="kd">function</span><span class="p">()</span> <span class="p">{</span> <span class="kd">var</span> <span class="nx">count</span> <span class="o">=</span> <span class="mi">0</span><span class="p">;</span> <span class="k">return</span> <span class="p">{</span> <span class="na">incr</span><span class="p">:</span> <span class="kd">function</span><span class="p">(){</span><span class="nx">count</span><span class="o">++</span><span class="p">;},</span> <span class="na">decr</span><span class="p">:</span> <span class="kd">function</span><span class="p">(){</span><span class="nx">count</span><span class="o">--</span><span class="p">;},</span> <span class="na">disp</span><span class="p">:</span> <span class="kd">function</span><span class="p">(){</span><span class="nx">console</span><span class="p">.</span><span class="nx">log</span><span class="p">(</span><span class="nx">count</span><span class="p">)}</span> <span class="p">}</span> <span class="p">}</span> </code></pre> </div> <p>I exposed three functions: <code>incr</code>, <code>decr</code>, <code>disp</code>, they operate private <code>count</code> variable, on the outer side, callee don't know <code>count</code>, callee can call <code>incr</code>, <code>decr</code>, <code>disp</code>, three functions are interfaces of <code>counter</code>.</p> <p>Another implementation of <code>counter</code> is expose <code>count</code>, callee increase count, decrease count, display count by operate <code>count</code> variable directly.</p> <p>When <code>incr</code>, <code>decr</code> logic is changed, e.g. increase count by 2, not 1. Interfaces are same as before, callee don't need change any codes.</p> oop TianZhen Wed, 02 Aug 2017 14:33:14 +0000 https://dev.to/dawncold/upgrade-x220-bios-to-143-mod https://dev.to/dawncold/upgrade-x220-bios-to-143-mod <p>original BIOS has some limition: RAM freq, hardware whitelist, missing 'Advanced' BIOS options, etc.</p> <p>I downloaded latest BIOS with modification from <a href="http://x220.mcdonnelltech.com/resources/">here</a>, if you want upgrade your X220 with faster RAM and 5G wifi(7620AC), you need upgrade your bios first.</p> <p>If you have a running Windows, you can save some time of creating WINPE USB stick, you should follow BIOS upgrade README.txt, if not, you need create a WINPE USB stick, you can run BIOS updater program on WINPE.</p> <p>After upgraded BIOS, I can't boot my Ubuntu system, I found a solution about boot loader missing after upgrade bios, <a href="https://askubuntu.com/questions/629734/will-updating-the-bios-remove-my-ubuntu-efi-entry">here</a>, what should be noticed is, you must use UEFI mode, not legacy mode, so check your BIOS boot mode, select <code>UEFI only</code>, and run in LIVE terminal:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code>sudo efibootmgr -c -L Grub -l /EFI/ubuntu/grubx64.efi </code></pre> </div> TianZhen Tue, 20 Jun 2017 16:03:34 +0000 https://dev.to/dawncold/install-debian-9-on-virtualbox https://dev.to/dawncold/install-debian-9-on-virtualbox <h1> why virtualbox </h1> <p><strong>UPDATE: The list of VMWare OS version only help you set basic resource (CPU, Memory, etc), so you can choose anyone and modify them on your need</strong></p> <p>I don't know which OS version should be chosen from VMWare OS version list. There are some Debian series, latest is Debian 8.x, but I need 9, also there is no Kernel 4.x item on the list, so I install virtualbox, I find a Debian (64bit).</p> <h1> disk space </h1> <p>It's better to generate a 15Gib or more disk. I use a 10Gib disk and got "no more space" error during installing GNOME.</p> <h1> additional software </h1> <p>compile virtualbox additional software needs some build softwares, you can get them from <code>build-essential</code> package, and build kernel modules need kernel header files, you can get them from <code>linux-headers-4.9.0-3-amd64</code>, which is default kernel version.<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code>sudo apt install build-essential linux-headers-4.9.0-3-amd64 </code></pre> </div> debian virtualbox linux TianZhen Tue, 06 Jun 2017 15:32:00 +0000 https://dev.to/dawncold/vmware-tools- https://dev.to/dawncold/vmware-tools- vmware TianZhen Sun, 12 Mar 2017 05:12:02 +0000 https://dev.to/dawncold/enlarge-ubuntu-root-partition-in-vmware https://dev.to/dawncold/enlarge-ubuntu-root-partition-in-vmware <p>I have a ubuntu vmware guest with 30Gb root partition, but it's not enough after restore data from production environment, so I have to enlarge it.</p> <ul> <li>shutdown ubuntu</li> <li>enlarge partition on vmware guest setting</li> <li>start ubuntu, it's better use text mode</li> <li> <code>sudo fdisk /dev/sda</code> (I have only one device in ubuntu, <code>sda</code>)</li> <li>delete swap partition and root partition with <code>d</code> command (yes, delete all partitions)</li> <li>create root partition (primary partition) with <code>n</code> command, be careful the start selector must be as same as old root partition (default is good), end selector is the size what you want the partition be, e.g. <code>+45G</code>, is a 45Gb partiton</li> <li>(optional) create swap partition if you need, with <code>n</code> command, and then change swap partition type with <code>t</code> command, swap type code is <code>82</code> </li> <li>write changes with <code>w</code> </li> <li>reboot ubuntu</li> <li> <code>sudo resize2fs /dev/sda1</code>, enlarge <code>sda1</code> to default new size, e.g. 45Gb</li> <li>(optional) <code>sudo mkswp /dev/sda2</code> if you create swap partition</li> <li>(optional) change swap partition record in <code>/etc/fstab</code>, new swap partition uuid or label could be found on <code>sudo blkid</code> </li> </ul> ubuntu vmware resize2fs TianZhen Sat, 11 Mar 2017 13:59:03 +0000 https://dev.to/dawncold/nas- https://dev.to/dawncold/nas- nas freenas ecc zfs TianZhen Wed, 15 Feb 2017 07:51:24 +0000 https://dev.to/dawncold/tcp-finwait2-timewait-and-closewait https://dev.to/dawncold/tcp-finwait2-timewait-and-closewait <p>Serverå’ŒClient都可以关闭已建立的连接，有可能Server先关闭，有可能Client先关闭，有可能同时关闭。同时关闭不太常见，Server关闭也不太常见（Server觉得有些Clientä¸€ç›´ä¿æŒç€è¿žæŽ¥å ç”¨èµ„æºï¼Œå…³é—­äº†æŸäº›Client的连接，一般来说Server是做服务的，不会主动断开，除非是某些特别的需要），Client先关闭比较常见。</p> <p>Client先关闭的情况：<a href="https://benohead.com/tcp-about-fin_wait_2-time_wait-and-close_wait/">https://benohead.com/tcp-about-fin_wait_2-time_wait-and-close_wait/</a></p> <p>Server先关闭的情况：<a href="https://benohead.com/c-understanding-close_wait-and-fin_wait_2/">https://benohead.com/c-understanding-close_wait-and-fin_wait_2/</a></p> <p>同时关闭连接的情况：<a href="http://www.tcpipguide.com/free/t_TCPConnectionTermination-4.htm">http://www.tcpipguide.com/free/t_TCPConnectionTermination-4.htm</a></p> TianZhen Thu, 09 Feb 2017 14:40:28 +0000 https://dev.to/dawncold/speed-up-your-pyres-worker https://dev.to/dawncold/speed-up-your-pyres-worker <p>We have a queue system for async works, it is build by <code>Pyres</code> with <code>Redis</code>. It works fine if your application is small, pyres worker forks a new process before doing job, and then terminate working process, worker just waits for new job.</p> <p>With out application is more and more complex, every new working process spends much time on loading, the package must be loaded, the package's dependencies must be loaded, e.g. working process only spends 1s on executing job code, but it needs 10s for loading.</p> <p>We think if <code>worker process</code> could load everything it needs, new process don't need load because of new process is forked by <code>worker process</code>, python knows which module is loaded, if you load a loaded module, it just returns, very fast. Maybe <code>Worker</code> class can be extened and implement <code>before_fork</code> method for loading something, loaded modules are available for new forked process.</p> <p>Another methods is that, if <code>working process</code> could exists for a long time, it don't need load modules again for some time. There is a <code>pyres_manager</code>, it creates a <code>manager</code>, <code>manager</code> creates some <code>minion</code>, <code>manager</code> put <code>minion</code> into <code>pool</code>. If you want to use <code>pyres_manager</code>, don't expect <code>resweb</code> works fine, it can't display <code>worker</code> because it don't know <code>minion</code>.</p> python pyres