Troubleshoot Autopilot Error Messages

Issue

Autopilot will fail if there are uncommitted SFTP code changes in the Dev environment.

Diagnosis

If commits are pushed to Dev after the Autopilot Multidev is created, you run the risk of having a merge conflict when Autopilot is pushed back to Dev.

If Autopilot attempted to deploy, then either: tests have already passed; or the less likely event that a test that was waiting, was approved after changing the Dev environment.

Solution

If you experience this error, it will likely be at the end of the flow, at which point you should re-run Autopilot with the latest changes to ensure all is covered in testing. Autopilot will also log an error, which can be found in the activity feed.

Pantheon's Autopilot engineers investigate each of these errors as they occur. Please contact Support via chat or ticket.

Issue

An error occurred because your site is currently running an unsupported version of Drush.

Diagnosis

If your Drupal site is not managed with Composer, Autopilot only works on sites that are running Drush 8. Composer-managed sites may use any compatible Drush version and will not display the Drush version error.

Autopilot only supports Integrated Composer; Build Tools sites cannot be updated.

Solution

To resolve this issue, switch to Drush 8 in the pantheon.yml file or use Integrated Composer.

Issue

• WordPress: This error message is displayed if a WordPress plugin or theme is not found after updates were attempted.

• Drupal: This error message is displayed if a Drupal module or theme update could not be applied.

Diagnosis

To determine which plugin is causing the issue:

1. Check the uncommitted changes for Autopilot Multidev. It is likely a plugin has been deleted and another plugin has been added.

2. Compare the list of plugins between the Autopilot and Dev environments and determine if there are any discrepancies. Ensure no entries are missing. If there is variance between the lists, such as a plugin that has been omitted or not updated on one of the lists, the plugin will be disabled on the Autopilot branch.

If you need assistance, contact Support.

Solution

Remove the plugin, theme, or module from the site if it is not being used. You can also revert the plugin, theme, or module to the original name.

Alternatively, you can add the plugin, theme, or module to the Excluded Updates list in Autopilot settings.

Pantheon's Autopilot engineers investigate each of these errors as they occur. Please contact Support via chat or ticket.

Issue

Applying extension updates failed.

Diagnosis

Applying updates with Drush or WP-CLI failed. This could be because of a read/write error, PHP fatal error thrown by the CMS, or another site-level error.

Solution

Ensure Drush or the WP-CLI works as expected on the Dev and Multidev environments.

Apply the plugin or theme updates either in the UI or with Drush or WP-CLI using Terminus. Resolve any issues or errors that arise. After conflicts are resolved, the updates should be committed to the master branch.

If possible, manually edit or update the plugins required to resolve the update failure and allow Autopilot to run its subsequent round of updates.

Issue

Applying upstream updates failed and changes cannot be automatically merged by Git.

Diagnosis

This error message most likely results from a merge conflict when applying upstream updates to the site. For sites that use Composer, this could be caused by a failed Composer build.

Solution

Resolve conflicts to apply updates. Use the auto-resolve option in the Dashboard to resolve conflicts in favor of the upstream Git repository. Alternatively, you can manually pull changes using Git, resolve the conflicts, and then push the updates to your Pantheon site. For more information, refer to the documentation on Applying Upstream Updates.

If the error is displayed due to a failed Composer build, use git diff to view changes, and examine the error in the log. Composer build logs are only available after the action completes or fails. For more information, refer to the documentation on Troubleshooting Code Syncs and Upstream Updates and Adding Dependencies to Your Upstream.

This error can also display if some of the updates failed testing. This results in partial updates, in which the updates that passed testing are applied, while the updates that failed testing are not applied. Refer to the Failing Tests section for more information on how to resolve failed tests.

Issue

The visual regression test (VRT) could not be completed due to a redirect error on one or more pages.

Diagnosis

Autopilot fails when a VRT page redirects more than 8 times or is stuck in an infinite loop.

Solution

To resolve this issue, remove the page from VRT settings or fix the redirect in the Dev environment for the page.

Issue

There are no visual regression testing pages in the Autopilot settings.

Solution

Add at least one site to the Autopilot VRT settings.

Issue

Autopilot failed to deploy to Test or Live; however, deploying to Dev from Multidev was successful. The most common reason for this is running clear cache or update db using Drush or the WP-CLI failed after the code was deployed. Ensure that clearing the cache using Drush or the WP-CLI works on the target environment.

Solution

If the Drush or WP-CLI steps failed during diagnosis, resolve any errors thrown by the CMS.

Run the deploy to Test or Live manually. Autopilot will also attempt to deploy again when the updates are tested, but immediately running updates again will likely result in the response that Autopilot has no new updates to apply, because the Dev environment already has changes.

If these steps fail, contact Support.

Issue

Failed to get a 200 OK response from the homepage of the Dev environment.

Solution

Ensure the Dev environment is live and reachable with no fatal errors and returns a 200 OK with curl or another check, for example curl -I https://dev-{SITE}.pantheonsite.io/.

CMS Error

Diagnosis

This can result from Drush or WP-CLI failing after db pull. This might be Autopilot specific, due to a site-level CMS issue, or could also be due to a platform-wide event.

Solution

Check that CLI cache clear steps work in the Dev environment. See if creating other Multidevs works correctly, delete the Autopilot environment and branch. Deleting the branch is important because the branch remains in Git if only the Multidev is deleted. If these actions works correctly, try running Autopilot again.

Composer Error

Diagnosis

The error could be caused by a Composer build failure if you are utilizing Integrated Composer.

Solution

Check the build log in the Site dashboard. Review the most recent commit to see if an error was posted.

You can also use Composer 2 to test locally and identify issues by running composer install or composer update

Issue

Updates could not be merged to the Dev environment.

Diagnosis

The most common reason for failure is due to a merge conflict. Changes were deployed in the Dev environment after Autopilot ran that resulted in a merge conflict. This error could also be because of a site-level PHP issue with clear-cache or update.php failing following a merge.

Solution

If failure is because of a merge conflict, run Autopilot again with latest changes.

If the issue is because of a PHP or CMS issue, resolve any issues causing clear-cache or update.php steps to fail, and manually merge again, or re-run Autopilot.

Diagnosis

The most common reason for clone content to fail is that drush/wp-cli clear cache failed at the end of a workflow.

Solution

Ensure the commands work. You will not be able to see if the workflow failed, but CSEs will be able to confirm if there was an issue with Drush or WP-CLI by replicating the step with Terminus.

Use the following steps:

• For WordPress:

  wp cache flush

• Drupal:

  drush cache-rebuild

Issue

Unsaved work in the development environment has prevented Autopilot from running.

Solution

Commit or discard the change, and run Autopilot again.

Issue

A CMS was not detected.

Solution

Install Drupal or WordPress on this site, and run Autopilot again. For more information on creating a new Drupal or WordPress site on Pantheon, refer to the documentation on Creating Sites.

Issue

A Sandbox site is frozen due to inactivity.

Solution

Unfreeze the site in the old dashboard before re-queueing updates. Note, unfreezing may take some time.

Diagnosis

Site is running a framework that is not supported. Supported frameworks are:

• Drupal with Integrated Composer
• Drupal 8 with Integrated Composer or Drush 8
• WordPress (note that WordPress Multisite (wordpress_network) is not currently supported)

Solution

Contact support for assistance.

Diagnosis

A WordPress Composer Managed site which is not using Composer is not supported by Autopilot.

Solution

Ensure that your WordPress Composer Managed site is properly configured. For additional help contact Support for assistance.

Diagnosis

There is an issue with Autopilot because Drush or WP-CLI did not work as expected in the Dev or Autopilot environment.

Run terminus remote:drush $SITE.dev -- pml or terminus remote:wp $SITE.dev -- plugin list and check for an unexpected output or errors. Alternatively, you can use the command $SITE.autopilot for the Autopilot environment.

Solution

After diagnosing the problem, resolve the issue that is causing the error on the Dev or Autopilot environment, then retry Autopilot. For additional help contact Support[/support] for assistance.

Issue

Quicksilver scripts failed to execute on the site.

Solution

Contact Support[/support] for assistance.

Issue

An issue with Autopilot occurred because the WordPress Dev, Test, or Live database could not be updated.

Solution

Update the database on the target environment, and then retry Autopilot. You will be prompted to run the update on the admin dashboard of the site, or you can run wp core update-db in the CLI. Contact Support[/link] for assistance if the issue persists.

Issue

This error is distinct from a merge conflict, and is encountered when there is no shared Git history between the site's commits and the upstream's commits. This error is most often encountered when the site's upstream has been changed, or when a different Git history has been force-pushed for the site's code repository.

Solution

Update the site's code or the upstream so that the site and upstream share a common history.

You can also disable upstream updates if you do not want Autopilot to maintain upstream updates, including core updates. Autopilot will continue to update plugins, themes, and modules.