mongodb MongoDB持续高CPU使用率
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/34951683/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
MongoDB constantly high cpu usage
提问by alsdkjasdlkja
According to docker container statistics, my mongo database consumes constantly between 250 and 350% cpu. That's pretty impressive since it's a single core system :P
根据 docker 容器统计，我的 mongo 数据库持续消耗 250% 到 350% 的 CPU。这是非常令人印象深刻的，因为它是一个单核系统：P
The sad part is that this is my production instance and much more sad is that it has to live until the next prod-backup and thats 3.5 more hours to go.
可悲的是，这是我的生产实例，更可悲的是它必须持续到下一次生产备份，而且还有 3.5 个小时。
I tried to do a mongotop but it tells me 0ms stats for all shown collections. Can I do anything else to figure out what's going on?
我试图做一个 mongotop，但它告诉我所有显示的集合的 0ms 统计数据。我能做些什么来弄清楚发生了什么吗？
PS: The db is up for 9 weeks and didn't cause problems.
PS：数据库已经运行了 9 周并且没有引起问题。
回答by Daniel F
There is a function called db.currentOp()which lists the currently running queries with very detailed information, it also includes the duration they have been running (secs_running).
有一个名为db.currentOp()的函数，它列出了当前正在运行的查询以及非常详细的信息，它还包括它们运行的持续时间 ( secs_running)。
You can then use the currentOp.opidwith db.killOp()to kill that query/operation.
然后您可以使用currentOp.opidwith db.killOp()来终止该查询/操作。
If db.currentOp()doesn't return any results, because there is no query which went havoc, then there's also db.setProfilingLevel()which will enable profiling by storing queries into the "local" database. Here's a "Tutorial" which is from the "M102: MongoDB for DBAs"Course.
如果db.currentOp()没有返回任何结果，因为没有查询造成严重破坏，那么还有db.setProfilingLevel()它将通过将查询存储到“本地”数据库来启用分析。这是来自“M102：MongoDB for DBA”课程的“教程” 。
Further information can also be found in this detailed article "Troubleshooting MongoDB 100% CPU load and slow queries"from Igor Khomenko.
还可以在Igor Khomenko 的这篇详细文章“对 MongoDB 100% CPU 负载和慢查询进行故障排除”中找到更多信息。
回答by Paulo Victor
The first and more important thing that you need to do is check your kind of queries, for example, in my case I had the same problem and when I checked my logs tail -f /var/log/mongodb/mongod.log(You can configure this logs in /etc/mongod.conf) I only saw the simple queries like db.brands.find({"field":"value"}) but I checked my indexes in "brands" collections and this field on queries was not indexed (db.brands.getIndexes()). The only thing that I did is index this field db.brands.ensureIndex({name:1},{unique:true}). Of course make sure if your field is unique because in this example I put as a unique. After that my CPU changed from 100% to 20%.
您需要做的第一件事也是更重要的事情是检查您的查询类型，例如，在我的情况下，我遇到了同样的问题，当我检查我的日志tail -f /var/log/mongodb/mongod.log（您可以在 中配置此日志/etc/mongod.conf）时，我只看到了简单的查询，例如db.brands.find({"field":"value"}) 但我检查了我在“brands”集合中的索引，并且查询中的这个字段没有被索引（db.brands.getIndexes()）。我所做的唯一一件事就是索引这个字段db.brands.ensureIndex({name:1},{unique:true})。当然要确保你的领域是独一无二的，因为在这个例子中我把它作为唯一的。之后，我的 CPU 从 100% 变为 20%。
So I am not saying that is your problem but could be, check your queries before doing some bigger things.
所以我并不是说这是你的问题，但可能是，在做一些更大的事情之前检查你的查询。
