Incident with Pull Requests, Git Operations and Issues

Frequent GitHub outages affecting pull requests, SSH-based git operations, Issues, and Actions are prompting developers and companies to question the platform’s reliability for critical workflows. Commenters share short‑term workarounds (such as switching to HTTPS, adding file caches, or decoupling deploys from GitHub Actions) and longer‑term strategies like moving to self‑hosted forges, alternative CI/CD systems, or GitLab. Some attribute the latest incident to a brownout tied to GitHub’s deprecation of older SSH key types and criticize the opacity and optimism of its official status reporting.

Nature of the Incident

  • Outage affected pull requests, git operations (push/pull/clone), issues, and later Actions.
  • SSH access in particular was intermittently failing; some could push once, then lost access.
  • HTTPS cloning/pulling often still worked, leading to confusion about scope.
  • Some observed multiple overlapping incidents listed on GitHub’s status site.

Impact on Workflows

  • Many developers briefly suspected compromised SSH keys or local misconfigurations.
  • Teams with deployments tightly coupled to GitHub Actions and GitHub Container Registry were blocked.
  • Even users with self-hosted CI (Jenkins, CircleCI, etc.) were impacted if source checkout depended on GitHub.
  • Some report push/pull reliability has been poor for a while, though others say they see no such issues, suggesting uneven behavior or local/network factors.

Causes and SSH Key Changes

  • Several commenters link the timing to GitHub’s scheduled “brownout” of old SSH algorithms (e.g., SHA‑1-based ssh-rsa and a specific DH group).
  • This is presented as a plausible cause, but acknowledged as an assumption, not confirmed.

Mitigations and Workarounds

  • Temporary switch from SSH to HTTPS via git config ... url."https://github.com/".insteadOf [email protected]:.
  • Using local/file caches for GitHub artifacts to shield builds from outages.
  • Designing deployments so that GitHub Actions only trigger scripts on a separate “deploy host,” which can be used manually during outages.
  • Simple bare repos with post-receive hooks and tools like gitolite suggested for access control.

Alternatives to GitHub / Actions

  • Strong interest in self-hosted or alternative forges: Gitea, Forgejo, GitLab, GitSocial, Codeberg (noting it’s F/LOSS-only), plus commercial services like Blacksmith and Buildkite.
  • Several run Forgejo/Gitea with their own runners; others favor self-hosted Jenkins.
  • Some argue self-hosting is worth the ops burden; others explicitly want managed services.

Status Page and Reliability Concerns

  • Many criticize GitHub’s status page as lagging reality or under-reporting, forcing teams to waste time debugging “green” systems that are actually broken.
  • Third-party monitors and “days since last incident” trackers are cited as more informative.
  • Some feel outages are now frequent enough that normal operation feels like the anomaly.
  • There is frustration that paying business users aren’t better isolated from usage spikes (e.g., AI-related load).

Meta / Cultural Commentary

  • Thread includes jokes about outages happening every day ending in “-day,” AI “fixing everything,” and calls to “give up GitHub.”
  • A few note the outage was relatively short, but recurring incidents are eroding trust in GitHub as core infrastructure.