I'm learning GraphQL and am using prisma-binding for GraphQL operations. I'm facing this nodemon error while I'm starting my Node.js server and its giving me the path of schema file which is auto generated by a graphql-cli. What is this error all about?
Error:
Internal watch failed: ENOSPC: System limit for number of file watchers reached, watch '/media/rehan-sattar/Development/All projects/GrpahQl/graph-ql-course/graphql-prisma/src/generated
There are some tips in the comments and I brought them to update this question. If you're facing this problem, then you are probably using a Linux distro and your project is hitting your system's file watchers limit.
Check your current inotify file watch limit by executing:
$ cat /proc/sys/fs/inotify/max_user_watches
You can set a new limit temporarily with:
$ sudo sysctl fs.inotify.max_user_watches=131070
$ sudo sysctl -p
Or you can set a permanent limit:
echo fs.inotify.max_user_watches= 131070 | sudo tee -a /etc/sysctl.conf && sudo sysctl -p
Here we are setting 131070 file limit (as suggested by @Dmitriy in the comments). If this limit doesn't works for your system you can twice this number.
This documentation provides more technical details.
If you are using Linux, your project is hitting your system's file watchers limit
To fix this, on your terminal, try:
echo fs.inotify.max_user_watches=524288 | sudo tee -a /etc/sysctl.conf && sudo sysctl -p
You need to increase the inotify watchers limit for users of your system. You can do this from the command line with:
sudo sysctl -w fs.inotify.max_user_watches=100000
That will persist only until you reboot, though. To make this permanent, add a file named /etc/sysctl.d/10-user-watches.conf with the following contents:
fs.inotify.max_user_watches = 100000
After making the above (or any other) change, you can reload the settings from all sysctl configuration files in /etc with sudo sysctl --system. (On older systems you may need to use sudo sysctl -p instead.)
I sometimes get this issue when working with Visual Studio Code on my Ubuntu machine.
In my case the following workaround helps:
Stop the watcher, close Visual Studio Code, start the watcher, and open Visual Studio Code again.
In order to test the changes, I temporary set the parameter with the value 524288.
sysctl -w fs.inotify.max_user_watches=524288
Then I proceed to validate:
npm run serve
And the problem was solved. In order to make it permanent, you should try to add a line in the file "/etc/sysctl.conf" and then restart the sysctl service:
cat /etc/sysctl.conf | tail -n 2
fs.inotify.max_user_watches=524288
sudo systemctl restart systemd-sysctl.service
I had the same problem. However, mine was coming from Webpack. Thankfully, they had a great solution on their site:
For some systems, watching many files can result in a lot of CPU or memory usage. It is possible to exclude a huge folder like node_modules using a regular expression:
module.exports = {
watchOptions: {
ignored: /node_modules/
}
};
This is a problem of inotify (inode notify) in the Linux kernel, so you can resolve it by using this command:
For a temporary solution until rebooting the pc, use the following command
sudo sysctl -w fs.inotify.max_user_watches=100000
A permanent solution: To make this permanent, add a file named /etc/sysctl.d/10-user-watches.conf with the following contents:
fs.inotify.max_user_watches = 10000
After making the change, reload the settings from all sysctl configuration files in /etc with sudo sysctl -p.
In my case, while I'm doing the nodemon command on the Linux server, I have my Visual Studio Code open (SSH to the server). So based on Juri Sinitson's answer, I just close Visual Studio Code and run the nodemon command again. And it works.
My nodemon command:
nodemon server.js via npm start
It can be hard to know how much to increase the number of watchers by. So, here's a utility to double the number of watchers:
function get_inode_watcher_count() {
find /proc/*/fd -user "$USER" -lname anon_inode:inotify -printf '%hinfo/%f\n' 2>/dev/null |
xargs cat |
grep -c '^inotify'
}
function set_inode_watchers() {
sudo sysctl -w fs.inotify.max_user_watches="$1"
}
function double_inode_watchers() {
watcher_count="$(get_inode_watcher_count)"
set_inode_watchers "$((watcher_count * 2))"
if test "$1" = "-p" || test "$1" = "--persist"; then
echo "fs.inotify.max_user_watches = $((watcher_count * 2))" > /etc/sysctl.d/10-user-watches.conf
fi
}
# Usage
double_inode_watchers
# to make the change persistent
double_inode_watchers --persist
I think most answers given here are correct, but using the systemctl command to restart my service solved the problem for me. Check the command below:
sudo systemctl restart systemd-sysctl.service
I use Vite alongside with Python and its venv module, and I encountered the same error because of a large venv folder (10k files)
Vite config allows ignoring directories from file watcher, so with venv ignored, the error was gone.
export default defineConfig({
server: {
watch: {
ignored: ['**/venv/**'],
}
},
})
You should follow answers such as this one:
Or:
And for latest Ubuntu versions, run sudo sysctl --system to read these settings anew.
However, in my case, my changes to these configuration files were not picked up, because I had already tweaked these settings a while ago... and forgot about it. And I had placed the conflicting configuration file in the wrong place.
According to man sysctl.d, these settings can be placed in /etc/sysctl.d/*.conf, /run/sysctl.d/*.conf and /usr/lib/sysctl.d/*.conf.
In my case I had two files:
/etc/sysctl.d/10-user-watches.conf
/usr/lib/sysctl.d/30-tracker.conf <<< Older file, with lower limit
Due to the naming convention, my older file was read last, and took precedence.
On Linux, I've actually run with sudo.
sudo npm start
