Until now, shipping a change meant holding every other change behind it. Lane changes let two deploys move at once, with a rollback that only touches the lane that went wrong.
What actually changed
Every deploy now gets its own lane and its own rollback point. If one fails, the other keeps moving. Nothing behind it is held.
- Two concurrent deploys per environment
- Per-lane rollback, not per-environment
- A held lane no longer blocks the queue
It is on for every workspace today. There is nothing to enable.
Per-lane rollback is exactly what we needed. Does it work with self-hosted runners?
It does — lanes are orchestrated on our side, so runners do not need an update.
Turned it on this morning (well, it was already on). First Friday deploy in months.