0bebb556
retro: what the volumes release cost
a73x 2026-09-05 18:00
Commit message
RETRO.md
| Old | New | ||
|---|---|---|---|
| @@ -71,3 +71,13 @@ One line per push to `main`: what slowed the work down. Enforced by | |||
| 71 | runs died at placement: a gate that dies does not reap its smoke VM, and the | 71 | runs died at placement: a gate that dies does not reap its smoke VM, and the |
| 72 | leftover plus a live test VM filled the devhost. Prove at the layer the code | 72 | leftover plus a live test VM filled the devhost. Prove at the layer the code |
| 73 | runs through, and check the host is empty before a run. | 73 | runs through, and check the host is empty before a run. |
| 74 | - Volumes release: the branch's own refactor broke the release. Ordering moved | ||
| 75 | to internal/version so the agent could reach it, and the ship script's tag | ||
| 76 | probe kept compiling against internal/server/release.Less — a symbol nothing | ||
| 77 | in the build graph references, so vet, lint and the boot gate all passed and | ||
| 78 | stage 1 of the stg ship found it instead. Scripts that compile Go inline are | ||
| 79 | outside every check the tree runs on itself. The rebase cost the same way | ||
| 80 | twice over: HardDeleteVM had gained a hostID argument on main, and a fake API | ||
| 81 | the new cloud-init test replaced wholesale stopped registering its VM, so the | ||
| 82 | first ci after a clean rebase failed in two places that had nothing to do | ||
| 83 | with volumes. | ||