Incident with Pull Requests, Issues, Git Operations and API Requests

Original Article ↗ Hacker News Discussion ↗

Frequent outages affecting GitHub pull requests, issues, git operations, and APIs are fueling frustration among developers and raising concerns about the platform’s reliability. Commenters point to surging load from AI-generated activity, Microsoft’s stewardship and Azure migration, and GitHub’s expanding feature surface as likely contributors, while noting that write-heavy workloads are inherently hard to scale. Many are now weighing alternatives such as self-hosted Forgejo, GitLab, or simpler git servers, arguing that centralized forges and free tiers may be increasingly unsustainable at current usage levels.

Reliability and Outage Patterns

  • Multiple users report recurring incidents: failed or delayed pushes, PR creation errors, missing commits in PR diffs, and broken or delayed Actions runs.
  • Several say outages feel “daily” or at least weekly in 2026, contrasting with years when GitHub downtime was rare enough to be surprising.
  • Status pages are criticized as under-reporting or visually hiding outages (all-green charts with tiny incident markers; incidents marked “resolved” while fallout continues).
  • Third‑party “honest” status dashboards are shared and seen as more aligned with user experience.

Perceived Causes: Scale, AI, Azure, Management

  • GitHub’s own stat of 14× year‑over‑year commit growth is cited; many assume write-heavy scaling is a core issue.
  • Some attribute problems to Microsoft: migration to Azure, AI mandates, and shifting GitHub into a broader “AI hype machine.”
  • Others argue AI is driving load via agents and “vibe coders,” overwhelming infrastructure and turning GitHub into a de facto DDoS target.
  • A minority push back, noting growth and distributed systems limits rather than malice; some suggest SEV definitions changed, making incidents look worse on paper.

Impact on Developers

  • People report lost time, aborted workdays, noisy Slack alerts, and business-side managers noticing slowdowns.
  • PR diffs not reflecting all commits are viewed as especially dangerous, risking incorrect merges.
  • Some say GitHub Actions slowness wastes paid CI minutes.

Alternatives to GitHub

  • GitLab opinions are mixed: stronger CI/CD but cluttered, ops‑centric UI; some call it a “mess,” others say differences vs GitHub are marginal.
  • Azure DevOps, Bitbucket, and TFS are mentioned, mostly with mild or negative sentiment.
  • Strong enthusiasm for Forgejo/Gitea/Codeberg/Tangled and plain self‑hosted Git; users praise speed, control, and avoiding GitHub’s limits.
  • Several have already migrated most projects off GitHub and encourage others to do so.

UI/UX and Tooling Preferences

  • Complaints about GitHub’s developer ergonomics (missing hashes, sluggishness, React rewrite).
  • Forgejo/Gitea are lauded for speed but seen as needing better visual design; themed instances (e.g., Blender) get praise.
  • A side thread debates LLM‑generated web UIs, design quality, and how much prompt quality matters.

Broader Reflections and Proposed Remedies

  • Some see a wider decline in software reliability (GitHub, cloud providers, other industries), linked to “just ship” culture and LLM-assisted development.
  • Suggested (often tongue‑in‑cheek) fixes: freeze new signups, rate‑limit or segregate AI/agent traffic, reintroduce strong QA, tie bonuses to uptime, or even “undo” the Microsoft acquisition.
  • Others emphasize that users can’t force structural changes; the realistic options are switching providers, self‑hosting, or accepting the current trajectory.