$ the-wire · showcase
CONSUL-ECS FIXES HEALTH-CHECK RACE CONDITION ACROSS AWS
By RepoJournal · Filed · About HashiCorp · Composed from the cited sources · methodology
2 people shipped this
Consul's ECS integration just eliminated a critical race condition that could leave service and proxy health checks diverged in production.
The consul-ecs team shipped an atomic batching fix [1] that consolidates three separate Catalog.Register calls into a single operation during health sync cycles. Previously, partial failures could desynchronize service-side and proxy-side dataplane checks, creating a silent failure mode where traffic routing decisions were based on stale health state. This is the kind of bug that doesn't surface in staging. Alongside this fix, both terraform-aws-consul-ecs [2] and consul-ecs [3] are bumping serf to 0.10.4 and consul-api to 1.34.4, addressing upstream dependency drift. The health-sync batching change is the story here; the dependency updates are table stakes. There's also a stageContainerHealthChecks refactor [4] landing in parallel to clean up the surrounding code path.
Action items
- → Review the atomic health-sync fix [ref:4] and plan deployment before next consul-ecs release hashicorp/consul-ecs [plan]
- → Merge dependency updates [ref:1] and [ref:3] to stay current on serf and consul-api hashicorp/terraform-aws-consul-ecs [plan]
References
- [1] health-sync: batch Consul health check updates into a single atomic Catalog.Register ↗ hashicorp/consul-ecs
- [2] chore(deps): update serf and consul-api versions ↗ hashicorp/terraform-aws-consul-ecs
- [3] chore(deps): update serf and consul-api versions ↗ hashicorp/consul-ecs
- [4] refactor stageContainerHealthChecks ↗ hashicorp/consul-ecs
