feat(cluster): Add Fleet autoscaling and notifications - #214
Merged
Merged
Conversation
![sourceant[bot]](https://avatars.githubusercontent.com/in/1161781?s=60&v=4){kind=small}
Administrators can select available orchestration and routing providers without being offered adapters that are not ready on the server.
The Swarm runtime now declares the Moby API contract it imports directly, keeping module validation reproducible in CI.
Each deployment can retain validated scaling limits and controller state across agent restarts, including on a single server without Fleet enabled.
Fleet can publish and remove validated replica routes through the existing Nginx manager, including configuration checks and reloads before traffic changes take effect.
Autoscaling evaluations now persist their state, execute safe decisions, and group blocked or failed actions into one deployment incident.
Server Info now reports the stable agent address used by Fleet. Before setup, it derives the address from the authenticated request so setup can offer an editable default.
Saved targets can be tested without exposing their connection URL. Responses identify the delivery type so clients can preserve credentials during edits.
Fleet operators can select K3s for stateless workload placement and scaling. Configured cluster credentials and namespaces are used for readiness checks and workload operations.
Fleet health transitions now open one node incident, suppress repeated outage alerts, and resolve the same incident when the node recovers. Related deployment failures remain grouped under that incident.
Fleet provider setup can now save and report the K3s cluster context required for readiness checks and workload operations.
Scaled replicas now use the existing deployment proxy configuration, retaining SSL, aliases, security controls, and request behavior. Removing managed routing restores the original Compose target.
Deployments can declare a portable scaling workload and storage policy. Operators receive specific compatibility blockers before Fleet activation can move or replicate the service.
Operators can save a scale-ready workload independently of other deployment metadata and immediately receive an updated compatibility assessment.
Compatibility responses now include the saved workload declaration so clients can edit storage and portability settings without reconstructing them.
Scale-ready services now retain inline environment, startup commands, working directories, health paths, and routed ports across Swarm and K3s. Inputs that cannot yet move safely remain activation blockers.
Scale-ready Swarm services now join the configured proxy overlay so the existing FlatRun proxy can reach replicas across Fleet nodes.
Scale-ready deployments can move traffic only after managed replicas are ready. Failed cutovers restore the original route and remove the incomplete workload.
Scale-ready deployments can cut over from Compose to Swarm through the authenticated deployment API. Runtime state remains available when policy evaluation is disabled.
Managed deployments retain their provider, service, replica count, and route across agent restarts so later reconciliation can safely update traffic.
Successful scale changes now retain the actual replica count and routable backends. Restarted agents no longer reconcile from stale scaling state.
Active managed workloads can now be reconciled on a fixed interval. Observation failures become one correlated incident per deployment instead of silent gaps.
Managed workloads now react to sustained container and host pressure on a fixed interval. Partial observations block scaling, and default placement remains on the local node until Fleet capacity is explicitly granted.
Fleet workloads stay on the local Swarm node unless a connected peer grants capacity. Consenting nodes receive an isolated placement label and enforce the configured replica ceiling.
K3s deployments now receive a stable Service, native Traefik ingress, and recurring utilization checks through the Kubernetes Metrics API.
Schema-driven clients can discover workload compatibility, activation, policy, and capacity claim operations from the agent API description.
Borrowed placement now requires declared workload limits and only uses peers whose CPU, memory, and replica grants can accommodate each replica.
Borrowed capacity now requires a matching runtime cluster and a live peer grant. Ready replicas enter routing after asynchronous startup, capacity credentials reach their dedicated claim operation, and activation state is durable before Compose stops.
Autoscaling now observes only managed replicas, preserves routing identity, and applies safe provider updates without unnecessary reloads. Fleet health events, capacity claims, incident fallback, advertised URLs, and request cancellation now follow their intended trust and lifecycle boundaries.
Managed activation now survives a disconnected request while retaining a bounded lifetime. Managed Nginx route copies can no longer mutate shared backend state outside synchronization.
nfebe
force-pushed
the
feat/cluster-capacity
branch
from
August 22, 2026 20:21
593ff74 to
5022062
Compare
Stored peers receive missing default policies automatically. Existing credentials adopt those policies without requiring reconnection.
The next beta includes Fleet management, managed scaling, and grouped notifications. Upgrade notes cover automatic repair for existing peer connections.
nfebe
force-pushed
the
feat/cluster-capacity
branch
from
August 22, 2026 21:37
5b8bf1c to
bc69954
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Turn connected FlatRun servers into one manageable Fleet, with scoped peer access, grouped infrastructure notifications, and autoscaling through Swarm or K3s.