https://blog.zhangjt.fun/series/tidb-%E6%95%85%E9%9A%9C%E6%8E%92%E6%9F%A5/Recent content in TiDB 故障排查 on DBA学习笔记Hugozh-CN281304173@qq.com (张天师)281304173@qq.com (张天师)Wed, 21 Oct 2026 16:28:23 +0800https://blog.zhangjt.fun/remember-the-problem-of-schedule-getting-stuck-z20fpdx/Thu, 28 Nov 2024 11:25:06 +0800281304173@qq.com (张天师)https://blog.zhangjt.fun/remember-the-problem-of-schedule-getting-stuck-z20fpdx/<h2 id="1-现象">1 现象 </h2> <p>业务反馈查询 TiFlash 报错：<code>No access path for table 'dwd_id_mapping' is found with 'tidb_isolation_read_engines' = 'tiflash,tidb', valid values can be 'tikv'.</code></p> <p>到集群中查看了一下，确实这个表的副本一直处于同步状态，但是进度一直没有发生变化。</p>https://blog.zhangjt.fun/tidb-jitter-troubleshooting-path-byxrq/Tue, 23 Jan 2024 15:24:54 +0800281304173@qq.com (张天师)https://blog.zhangjt.fun/tidb-jitter-troubleshooting-path-byxrq/<h2 id="1-select执行流程">1 select执行流程 </h2> <img class="td-image" src="https://blog.zhangjt.fun/images/Pasted-image-20220321114021-20261020175823-n2i1ljj.png" alt="Pasted image 20220321114021" loading="lazy" decoding="async"><h2 id="2-insert流程">2 INSERT流程 </h2> <img class="td-image" src="https://blog.zhangjt.fun/images/Pasted-image-20220321114035-20261020175841-wya4xan.png" alt="Pasted image 20220321114035" loading="lazy" decoding="async"><h3 id="21-落盘步骤">2.1 落盘步骤 </h3> <img class="td-image" src="https://blog.zhangjt.fun/images/Pasted-image-20220321115327-20261020175853-05iw1hq.png" alt="Pasted image 20220321115327" loading="lazy" decoding="async"><p>‍</p>https://blog.zhangjt.fun/tidb-high-cpu-usage/Tue, 24 May 2022 16:11:36 +0800281304173@qq.com (张天师)https://blog.zhangjt.fun/tidb-high-cpu-usage/<h2 id="背景">背景 </h2> <ul> <li>TiDB Server两个节点CPU使用率都比较高 <img src="https://blog.zhangjt.fun/images/posts/tidb-high-cpu-usage/cpu-usage-nodes.svg" alt="TiDB Server 两个节点 CPU 使用率" loading="lazy" decoding="async"></li> <li>查看QPS指标，可以看到14点到19点之间是基本没有流量的 <img src="https://blog.zhangjt.fun/images/posts/tidb-high-cpu-usage/qps-metrics.svg" alt="QPS 指标，14 点到 19 点基本无流量" loading="lazy" decoding="async"></li> <li>登陆机器可以发现，有很多连接其实已经失效，但是还残留服务器上： <img src="https://blog.zhangjt.fun/images/posts/tidb-high-cpu-usage/stale-connections.svg" alt="服务器上残留的失效连接" loading="lazy" decoding="async"></li> <li>怀疑Server负载高和这些诡异的SQL有关系，即便没有释放，也没办法KILL，但是还是占用了一个核的CPU <img src="https://blog.zhangjt.fun/images/posts/tidb-high-cpu-usage/2022-05-29-mxWOpSfh5PwAsbL.png" alt="20220529204849" loading="lazy" decoding="async"> <ul> <li><a href="https://github.com/pingcap/tidb/pull/29212">https://github.com/pingcap/tidb/pull/29212</a></li> <li><a href="https://github.com/pingcap/tidb/issues/24031">https://github.com/pingcap/tidb/issues/24031</a></li> </ul> </li> </ul> <h2 id="解决方案">解决方案 </h2> <ol> <li> <p>将流量转移到其他机器，可选途径有：</p>