Ops · Deploy
A build on a diet
Rewrote the ignore rules so every deploy ships only what the image needs.
- 54GB → 5GB
- build context
- 01
Symptom
Deploys to staging and production were slow, and the build machine kept running out of disk.
- 02
Root cause
The ignore rules missed local data, media and dependency folders, so every build sent about 54GB to Docker.
- 03
Fix
Audited what the image actually needs and rewrote the ignore rules to send only that.
- 04
Impact
Build context down from ~54GB to ~5GB, with faster deploys.
Debug log
- $measure the context size before touching anything
- $list the largest folders sent to the daemon
- $allow only what the image needs
- $re-measure on staging, then production
- ✓resolved
A thousand welcomes