Rolling vs Blue-Green for Zero-Downtime Meteor Migration
0 reputation · 13 Aug 2021, 23:11 UTC
0 reputation · 13 Aug 2021, 23:11 UTC
The goal is a zero-downtime migration for a small Meteor application.
The choice is between rolling deployment requiring dual-schema support or blue-green deployment necessitating duplicate infrastructure.
Meteor's DDP lifecycle lacks a built-in mechanism to gracefully drain active in-flight method calls during a traffic cutover.
35775 reputation · 14 Aug 2021, 06:29 UTC
Rolling deployment uses a single codebase with dual‑schema support, while blue‑green duplicates the whole environment to avoid schema conflicts.
The main trade‑off is operational simplicity versus isolation cost. Rolling keeps one server group, requires manual draining of active DDP method calls, and relies on schema version compatibility. Blue‑green runs two identical stacks, switches traffic at the load balancer, and eliminates schema‑drain concerns but adds duplicate infrastructure and higher expense.
Do you have a reliable endpoint or metric that reports active DDP connections at the moment of cutover? This information is needed to confirm that the manual drain step can be performed safely.
Use comments to ask for clarification. Post a solution as an answer.
No question comments on this page.