failed to create directory via template ‘/tmp/chrome.XXXXXXX’: No space left on device

Deployed, everything worked – but after some time I am seeing this:

@serverless-chrome/lambda: Error trying to spawn chrome: { Error: Command failed: mktemp -d -t chrome.XXXXXXX
mktemp: failed to create directory via template ‘/tmp/chrome.XXXXXXX’: No space left on device

at checkExecSyncError (child_process.js:481:13)
at Object.execSync (child_process.js:521:13)
at makeTempDir (/var/task/node_modules/@serverless-chrome/lambda/dist/bundle.cjs.js:39:24)
at Launcher.prepare (/var/task/node_modules/@serverless-chrome/lambda/dist/bundle.cjs.js:108:52)
at /var/task/node_modules/@serverless-chrome/lambda/dist/bundle.cjs.js:246:16
at throw (native)
at step (/var/task/node_modules/@serverless-chrome/lambda/dist/bundle.cjs.js:69:193)
at /var/task/node_modules/@serverless-chrome/lambda/dist/bundle.cjs.js:69:404
at process._tickDomainCallback (internal/process/next_tick.js:135:7)
error: null,
cmd: 'mktemp -d -t chrome.XXXXXXX',
file: '/bin/sh',
args: [ '/bin/sh', '-c', 'mktemp -d -t chrome.XXXXXXX' ],
options: 
{ shell: true,
file: '/bin/sh',
args: [ '/bin/sh', '-c', 'mktemp -d -t chrome.XXXXXXX' ],

Looks like some cleanup is not happening?

(The lambda is kept warm by frequent usage)

"@serverless-chrome/lambda@1.0.0-53":
  version "1.0.0-53"

EDIT: Hmm, I do see there is code that should perform that cleanup, but I guess something prevents it from being reached.

Random idea: rm -rf /tmp/chrome.??????? on startup? (instead of at the end of the lambda invocation)

[2803media]

Thanks @emilburzo for this info, Yesterday I added this to my function, I hope it will temporary solve this problem:

'use strict';
const puppeteer = require('puppeteer');

const exec = require('child_process').exec;

module.exports.index = async (event, context) => {
  
  exec('rm -r /tmp/core.* || true', function(err,stdout,stderr){     
   if(err)
   console.log('Directory Empty', err);
   else
   console.log("Files Deleted");
  });
  
  // some code

}

[CarolynWebster]

I ran into this issue as well after months of use. The above code from 2803media worked locally - so I will try it deployed and hopefully it will keep this issue at bay.