lablup / lablup/backend.ai

Deployment revision update causes downtime by terminating old route before new route is ready

Open
#11,110 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
670
Forks
183
Avg merge
17h 7m
Merged PRs (30d)
358

Description

## Background

When a deployment is updated (e.g., model service configuration change) and a new revision is created, the expected behavior is:

1. New revision's route is created and becomes ready
2. Old revision's route is terminated (blue-green style)

## Current behavior:

1. Old revision's route is terminated first
2. New revision's route is created afterward

This causes downtime during deployment updates — the service is unavailable between the old route termination and the new route becoming ready.

## Success Criteria

- [ ] Deployment update creates new revision route before terminating old revision route

- [ ] No downtime window during revision transition

- [ ] Rollback scenario (new revision fails to start) does not leave service without any active route

JIRA Issue: BA-5743

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.