Skip to main content

Troubleshooting GitHub Actions for your enterprise

Troubleshooting common issues that occur when using GitHub Actions on GitHub Enterprise Server.

Qui peut utiliser cette fonctionnalité ?

Site administrators can troubleshoot GitHub Actions issues and modify GitHub Enterprise Server configurations.

Checking the health of GitHub Actions

You can check the health of GitHub Actions on votre instance GitHub Enterprise Server with the ghe-actions-check command-line utility. For more information, see "Utilitaires de ligne de commande" and "Accès à l’interpréteur de commandes d’administration (SSH)."

Configuring self-hosted runners when using a self-signed certificate for GitHub Enterprise Server

Nous vous recommandons vivement de configurer TLS sur GitHub Enterprise Server avec un certificat signé par une autorité de confiance. Bien qu’un certificat autosigné puisse fonctionner, une configuration supplémentaire est nécessaire pour vos exécuteurs autohébergés. Elle n’est pas recommandée pour les environnements de production. For more information, see "Configuring TLS."

Installing the certificate on the runner machine

For a self-hosted runner to connect to a GitHub Enterprise Server using a self-signed certificate, you must install the certificate on the runner machine so that the connection is security hardened.

For the steps required to install a certificate, refer to the documentation for your runner's operating system.

Configuring Node.JS to use the certificate

Most actions are written in JavaScript and run using Node.js, which does not use the operating system certificate store. For the self-hosted runner application to use the certificate, you must set the NODE_EXTRA_CA_CERTS environment variable on the runner machine.

You can set the environment variable as a system environment variable, or declare it in a file called .env in the self-hosted runner application directory (that is, the directory into which you downloaded and unpacked the runner software).

For example:

NODE_EXTRA_CA_CERTS=/usr/share/ca-certificates/extra/mycertfile.crt

Environment variables are read when the self-hosted runner application starts, so you must set the environment variable before configuring or starting the self-hosted runner application. If your certificate configuration changes, you must restart the self-hosted runner application.

Configuring Docker containers to use the certificate

If you use Docker container actions or service containers in your workflows, you might also need to install the certificate in your Docker image in addition to setting the above environment variable.

Configuring HTTP proxy settings for GitHub Actions

Si vous avez un serveur proxy HTTP configuré sur votre instance GitHub Enterprise Server :

  • Vous devez ajouter .localhost, 127.0.0.1 et ::1 à la liste d’Exclusion du proxy HTTP (dans cet ordre).
  • Si votre emplacement de stockage externe n’est pas routable, vous devez également ajouter l’URL de votre stockage externe à la liste d’exclusion.

Pour plus d’informations sur la modification de vos paramètres de proxy, consultez « Configuring an outbound web proxy server ».

If these settings aren't correctly configured, you might receive errors like Resource unexpectedly moved to https://IP-ADDRESS when setting or changing your GitHub Actions configuration.

Runners not connecting to GitHub Enterprise Server with a new hostname

Avertissement : ne modifiez pas le nom d’hôte de GitHub Enterprise Server après la configuration initiale. La modification du nom d’hôte entraînera un comportement inattendu, pouvant aller jusqu’à des interruptions de l’instance et l’invalidation des clés de sécurité des utilisateurs. Si vous avez modifié le nom d’hôte de votre instance et rencontrez des problèmes, contactez Support GitHub Enterprise ou Support Premium GitHub.

If you deploy GitHub Enterprise Server in your environment with a new hostname and the old hostname no longer resolves to your instance, self-hosted runners will be unable to connect to the old hostname, and will not execute any jobs.

You will need to update the configuration of your self-hosted runners to use the new hostname for votre instance GitHub Enterprise Server. Each self-hosted runner will require one of the following procedures:

  • In the self-hosted runner application directory, edit the .runner and .credentials files to replace all mentions of the old hostname with the new hostname, then restart the self-hosted runner application.
  • Remove the runner from GitHub Enterprise Server using the UI, and re-add it. For more information, see "Suppression d’exécuteurs auto-hébergés" and "Ajout d’exécuteurs auto-hébergés."

Stuck jobs and GitHub Actions memory and CPU limits

GitHub Actions is composed of multiple services running on votre instance GitHub Enterprise Server. By default, these services are set up with default CPU and memory limits that should work for most instances. However, heavy users of GitHub Actions might need to adjust these settings.

You may be hitting the CPU or memory limits if you notice that jobs are not starting (even though there are idle runners), or if the job's progress is not updating or changing in the UI.

1. Check the overall CPU and memory usage in the management console

Access the management console and use the monitor dashboard to inspect the overall CPU and memory graphs under "System Health". For more information, see "Accès au tableau de bord moniteur."

If the overall "System Health" CPU usage is close to 100%, or there is no free memory left, then votre instance GitHub Enterprise Server is running at capacity and needs to be scaled up. For more information, see "Augmentation des ressources processeur ou mémoire."

2. Check the Nomad Jobs CPU and memory usage in the management console

If the overall "System Health" CPU and memory usage is OK, scroll down the monitor dashboard page to the "Nomad Jobs" section, and look at the "CPU Percent Value" and "Memory Usage" graphs.

Each plot in these graphs corresponds to one service. For GitHub Actions services, look for:

  • mps_frontend
  • mps_backend
  • token_frontend
  • token_backend
  • actions_frontend
  • actions_backend

If any of these services are at or near 100% CPU utilization, or the memory is near their limit (2 GB by default), then the resource allocation for these services might need increasing. Take note of which of the above services are at or near their limit.

3. Increase the resource allocation for services at their limit

  1. Log in to the administrative shell using SSH. For more information, see "Accès à l’interpréteur de commandes d’administration (SSH)."

  2. Run the following command to see what resources are available for allocation:

    nomad node status -self
    

    In the output, find the "Allocated Resources" section. It looks similar to the following example:

    Allocated Resources
    CPU              Memory          Disk
    7740/49600 MHZ   23 GiB/32 GiB   4.4 GiB/7.9 GiB
    

    For CPU and memory, this shows how much is allocated to the total of all services (the left value) and how much is available (the right value). In the example above, there is 23 GiB of memory allocated out of 32 GiB total. This means there is 9 GiB of memory available for allocation.

    Warning: Be careful not to allocate more than the total available resources, or services will fail to start.

  3. Change directory to /etc/consul-templates/etc/nomad-jobs/actions:

    cd /etc/consul-templates/etc/nomad-jobs/actions
    

    In this directory there are three files that correspond to the GitHub Actions services from above:

    • mps.hcl.ctmpl
    • token.hcl.ctmpl
    • actions.hcl.ctmpl
  4. For the services that you identified that need adjustment, open the corresponding file and locate the resources group that looks like the following:

    resources {
      cpu = 512
      memory = 2048
      network {
        port "http" { }
      }
    }
    

    The values are in MHz for CPU resources, and MB for memory resources.

    For example, to increase the resource limits in the above example to 1 GHz for the CPU and 4 GB of memory, change it to:

    resources {
      cpu = 1024
      memory = 4096
      network {
        port "http" { }
      }
    }
    
  5. Save and exit the file.

  6. Run ghe-config-apply to apply the changes.

    When running ghe-config-apply, if you see output like Failed to run nomad job '/etc/nomad-jobs/<name>.hcl', then the change has likely over-allocated CPU or memory resources. If this happens, edit the configuration files again and lower the allocated CPU or memory, then re-run ghe-config-apply.

  7. After the configuration is applied, run ghe-actions-check to verify that the GitHub Actions services are operational.

Troubleshooting failures when Dependabot triggers existing workflows

Après avoir configuré les mises à jour Dependabot pour votre instance GitHub Enterprise Server, il se peut que constatiez des défaillances quand des workflows existants sont déclenchés par des événements Dependabot.

Par défaut, les exécutions de workflows GitHub Actions qui sont déclenchées par Dependabot à partir d’événements push, pull_request, pull_request_review ou pull_request_review_comment ont considérées comme ayant été ouvertes à partir d’une duplication (fork) de dépôt. Contrairement aux workflows déclenchés par d’autres acteurs, cela signifie qu’ils reçoivent un GITHUB_TOKEN en lecture seule et qu’ils n’ont pas accès aux secrets qui sont normalement disponibles. Ainsi, les flux de travail qui tentent d’écrire dans le dépôt échouent quand ils sont déclenchés par Dependabot.

Il existe trois façons de résoudre ce problème :

  1. Vous pouvez mettre à jour vos workflows de sorte qu’ils ne soient plus déclenchés par Dependabot en utilisant une expression de ce type : if: github.actor != 'dependabot[bot]'. Pour plus d’informations, consultez « Expressions ».
  2. Vous pouvez modifier vos workflows de sorte qu’ils utilisent un processus en deux étapes qui comprend pull_request_target qui ne présente pas ces limites. Pour plus d’informations, consultez « Automatisation de Dependabot avec GitHub Actions ».
  3. Vous pouvez permettre aux workflows déclenchés par Dependabot d’accéder aux secrets et autoriser le terme permissions à augmenter l’étendue par défaut de GITHUB_TOKEN. For more information, see "Providing workflows triggered by Dependabot access to secrets and increased permissions" below.

Providing workflows triggered by Dependabot access to secrets and increased permissions

  1. Log in to the administrative shell using SSH. For more information, see "Accès à l’interpréteur de commandes d’administration (SSH)."

  2. To remove the limitations on workflows triggered by Dependabot on votre instance GitHub Enterprise Server, use the following command.

    ghe-config app.actions.disable-dependabot-enforcement true
    
  3. Apply the configuration.

    ghe-config-apply
    
  4. Return to GitHub Enterprise Server.

Troubleshooting bundled actions in GitHub Actions

If you receive the following error when installing GitHub Actions in GitHub Enterprise Server, you can resolve the problem by installing the official bundled actions and starter workflows.

A part of the Actions setup had problems and needs an administrator to resolve.

To install the official bundled actions and starter workflows within a designated organization in GitHub Enterprise Server, follow this procedure.

  1. Identify an organization that will store the official bundled actions and starter workflows. You can create a new organization or reuse an existing one.

  2. Log in to the administrative shell using SSH. For more information, see "Accès à l’interpréteur de commandes d’administration (SSH)."

  3. To designate your organization as the location to store the bundled actions, use the ghe-config command, replacing ORGANIZATION with the name of your organization.

    ghe-config app.actions.actions-org ORGANIZATION
    

    and:

    ghe-config app.actions.github-org ORGANIZATION
    
  4. To add the bundled actions to your organization, unset the SHA.

    ghe-config --unset 'app.actions.actions-repos-sha1sum'
    
  5. Apply the configuration.

    ghe-config-apply
    

After you've completed these steps, you can resume configuring GitHub Actions at "Getting started with GitHub Actions for GitHub Enterprise Server."