Migrating 1TB to GitHub: Lessons Learned from the Field
summary
Migrating ~1TB of repos requires recreating history commit‑by‑commit (change‑list → Git commits), writing scripts (Copilot helps), and stripping giant binaries from history then converting them to Git LFS pointers; common surprises include a single repo of ~100GB and repositories holding Photoshop installs and marketing videos. Perforce migrations use Git P4 to clone depots but still need custom replay logic; pipeline migration is the slowest part, often solved by a mirror pipelines phase. Expect long discovery, identity remapping, and rate/size limits to bite.
AI-written summary · about
chapters (12)
- 0:00Introduction and setup
- 3:51Interactive poll on Git knowledge
- 7:06Why migrate to GitHub
- 12:00Understanding Git internals
- 21:38Git LFS and binary files
- 24:45Migration from Perforce
- 31:19Migration from Azure DevOps
- 33:46Challenges with pipelines
- 37:57Discovery phase importance
- 40:45Post-migration validation
- 42:09Key takeaways and advice
- 43:14Q&A session
related talks
42mTracking Huge Files with Git LFS
27mGitLFS - How to handle large files in Git
22mHow We Migrated Thousands of Repositories to SaaS With Developer Experience
58mGit at Scale
21mFighting Git Workflows
35mTerrible Ideas in Git
28mLarge scale refactoring: from start to finish
17mProject Immortality: Using GitHub To Make Your Work Live Forever