Skip to main content

Overloaded Workers

Learn more about overloaded workers on Pantheon.

This section provides information on how to resolve overloaded workers on Pantheon.

Overloaded Workers

It's possible that pages will timeout before they are picked up by the back-end if your PHP workers are overloaded. This can happen if you are suddenly hit with a flood of un-cachable or authenticated traffic.

2014/05/15 08:57:21 [error] 31914#0: *13543 connect() to unix:/srv/bindings/0181b7c2caqe34534qw34533453e69cd027b13556df00/run/php-fpm.sock failed (11: Resource temporarily unavailable) while connecting to upstream, client:, server: , request: "GET /index.php?q=user HTTP/1.0", upstream: "fastcgi://unix:/srv/bindings/0181b7c2caqe34534qw34533453e69cd027b13556df00/run/php-fpm.sock:", host: ""
2014/05/15 08:57:21 [alert] 31914#0: *13546 128 worker_connections are not enough while connecting to upstream, client: unix:ded-fo, server: , request: "GET /index.php?q=user HTTP/1.0", upstream: "", host: ""

Fix Errors

Your site's performance will deteriorate as it uses its resources to log errors to your disk if your site is throwing a lot of warnings or notices. This can slow down your site by performing database write operations. The solution is not to disable logging but to fix the errors.

Even with logging disabled, these errors will still be written to the server PHP error logs, which means you should be resolve the errors as soon as possible.

Optimize the Site

Long running processes can max out backend resources on your site, including:

  • Batch jobs
  • Background tasks
  • Heavy operation cron jobs

Use New Relic® Performance Monitoring to identify performance bottlenecks, fix errors, and make changes to enhance performance.

Upgrade Your Plan

Consider upgrading your plan for more resources after all the errors have been resolved and the views, batches, and tasks have been optimized. Select the most appropriate plan for the resource usage of your site.

Unexpected Timeouts

Buggy code can cause wide-ranging issues. Bugs ranging from Drupal running cron on every page load, to the Drupal module advanced_help spidering the entire code tree looking for help files can cause slow page load times that trigger timeouts.

Debugging with New Relic® Performance Monitoring or looking at your php slow logs can help you determine why timeouts are occurring in unexpected places.

More Resources