Visual Studio Code is unable to watch for file changes in this large workspace

2020-12-04 16:16:20

参考地址 1 Visual Studio Code on Linux

参考地址2  How Meteor watches your files for changes

解决办法   直接输入

echo fs.inotify.max_user_watches=524288 | sudo tee -a /etc/sysctl.conf && sudo sysctl -p


When you run your Meteor app in local development mode, the meteor command-line tool watches your source code for changes. (This page describes the behavior as of Meteor 1.0.2.1.)

There are two different strategies that meteor uses for change watching: efficient system-specific APIs like inotify (on Linux) and kqueue (on OSX), and a slower but more reliable "stat polling" method. inotify/kqueue is less reliable because it does not work on every system, and sometimes meteor can't even tell if it doesn't work. Stat polling is slower in two ways: it only looks for changes once every "polling interval", and the actual check that it does consumes more CPU.

By default, meteor uses both inotify/kqueue and stat polling, with a polling interval of 5 seconds; if it detects that inotify/kqueue isn't working, it uses a polling interval of 0.5 seconds.

This means that on most systems, changes are detected very quickly by inotify/kqueue. On systems where inotify/kqueue doesn't work, the stat polling fallback still works, but it does use more CPU than if inotify/kqueue works; and if meteor was unable to automatically detect that inotify/kqueue didn't work, you may have to wait up to 5 seconds for stat polling to notice file changes.

The rest of this wiki page describes two ways in which you can improve the performance of Meteor file watching.

  • There are some circumstances where inotify doesn't work (causing meteor to fall back to the CPU-hungry 0.5 second poll), but a simple configuration change will make inotify work.

  • There are some circumstances where inotify/kqueue doesn't work but meteor is unable to detect this, so only the high-latency 5 second poll happens. While you can't make inotify/kqueue work in this case, you can at least tell meteor to use a faster poll to detect changes sooner.

Making inotify work on Linux systems where it doesn't work

(Note: this section is Linux-specific: it will not work on OS X, and the warning message below should not be displayed there. If you do see this warning on OS X, please file a bug!)

Linux users may see a message saying:

=> It looks like a simple tweak to your system's configuration will make many
   tools (including this Meteor command) more efficient. To learn more, see
   https://github.com/meteor/meteor/wiki/File-Change-Watcher-Efficiency

This message means that meteor tried to use the Linux inotify API to watch for file changes, but you've already reached the system-wide limit for the number of files that you can watch (probably in a different program entirely!).

Fortunately, this is easy to fix. Just run this command at your shell, entering your password when prompted:

echo fs.inotify.max_user_watches=524288 | sudo tee -a /etc/sysctl.conf && sudo sysctl -p

(Alternatively, use another mechanism to edit /etc/sysctl.conf as root and add the line fs.inotify.max_user_watches=524288 to the bottom, then run sudo sysctl -p.)

After running these commands, you should no longer see the "simple tweak" message when running meteor. (If you still do, file a bug!) Meteor is now using the efficient inotify API

If you do not have superuser access on your computer, you won't be able to run these commands. In that case, you may set the METEOR_WATCH_FORCE_POLLING environment variable as described in the next section, which will hide the warning and not try to use inotify at all.

Running meteor on a network/shared filesystem

Some filesystems don't support inotify/kqueue at all, but also don't allow meteor to detect this error case. In this situation, meteor will still detect file changes with the stat poller at the 5 second polling interval, but you may want to poll more often in order to detect changes faster.

The most common filesystems with this issue are network or shared filesystems like NFS or VirtualBox shared folders (vboxsf). For example, if you are using Vagrant or a similar tool to run meteor inside a virtual machine, and you are sharing your app source directory between your host machine and the VM, inotify may be unable to detect changes made on your host machine. The symptom will be that meteor doesn't even start trying to rebuild your app until up to 5 seconds after you make the change.

To improve this situation, on the machine where you are running meteor (eg, inside your VM), set the METEOR_WATCH_FORCE_POLLING environment variable to any non-empty value. For example, put the line METEOR_WATCH_FORCE_POLLING=true into the .bashrc file in your home directory.

If you would like finer control over the polling interval (to put it somewhere between the options of 5 seconds and 500 ms), set the METEOR_WATCH_POLLING_INTERVAL_MS environment variable to a number.


  • 2019-08-27 05:43:13

    Laravel 门面自动补全工具 laravel-ide-helper

    当我们在 PhpStorm 编辑器中,开发 Laravel 框架的项目时,很多类方法都不能自动补全和定位,比如 Facade 门面的方法,DB::table()、Route::get() 等。

  • 2019-08-28 08:28:36

    Js apply,call方法详解,及其apply()方法的妙用

    在给对象参数的情况下,如果参数的形式是数组的时候,比如apply示例里面传递了参数arguments,这个参数是数组类型,并且在调用Person的时候参数的列表是对应一致的(也就是Person和Student的参数列表前两位是一致的) 就可以采用 apply , 如果我的Person的参数列表是这样的(age,name),而Student的参数列表是(name,age,grade),这样就可以用call来实现了,也就是直接指定参数列表对应值的位置(Person.call(this,age,name,grade));

  • 2019-08-28 08:49:47

    Nginx + fastcgi + php 的原理与关系

    CGI对每个请求会parse一遍对应脚本的配置文件(如php.ini), 加载配置和扩展,初始化执行环境,性能非常差,所有有了下面的流程:

  • 2019-08-28 09:23:15

    php单例模式

    单例模式,是一种常见的软件设计模式。在应用这个模式时,单例对象的类必须保证只有一个实例存在。

  • 2019-08-28 22:45:02

    彻底搞懂Gradle、Gradle Wrapper与Android Plugin for Gradle的区别和联系

    Gradle是个构建系统,能够简化你的编译、打包、测试过程。熟悉Java的同学,可以把Gradle类比成Maven。Gradle Wrapper的作用是简化Gradle本身的安装、部署。不同版本的项目可能需要不同版本的Gradle,手工部署的话比较麻烦,而且可能产生冲突,所以需要Gradle Wrapper帮你搞定这些事情。Gradle Wrapper是Gradle项目的一部分。