Files
PandeyandGitHub bebaa42fd4 feat(factory): add service state tracking, AwaitHealthy, depends_on, and /healthz (#10671)
* feat(factory): add service state tracking, AwaitHealthy, depends_on, and /healthz endpoint

Add explicit lifecycle state tracking to factory.Registry services
(starting/running/failed) modeled after Guava's ServiceManager. Services
can declare dependencies via NewNamedService(..., dependsOn) which are
validated for unknown refs and cycles at registry creation. AwaitHealthy
blocks until all services reach running state. A /healthz endpoint is
wired through signozapiserver returning 200/503 with per-service state.

* feat(apiserver): move health endpoints to /api/v2/ and register readyz, livez

* refactor(factory): use gonum for cycle detection, return error on cycles, fix test assertions

Replace custom DFS cycle detection with gonum's topo.Sort + TarjanSCC.
Dependency cycles now return an error from NewRegistry instead of being
silently dropped. Use assert for final test assertions and require only
for intermediate setup errors.

* chore: go mod tidy

* refactor(factory): decouple Handler from Registry, wire through Handlers struct

Move Handler implementation to a private handler struct with NewHandler
constructor instead of methods on *Registry. Route handler through the
existing Handlers struct as RegistryHandler. Rename healthz.go to
registry.go in signozapiserver. Fix handler_test.go for new param.

* feat(factory): add ServiceWithHealthy interface, add Healthy to authz, user depends on authz

Add ServiceWithHealthy interface embedding Service + Healthy. NamedService
now delegates Healthy() to the underlying service, eliminating unwrapService.
AuthZ interface requires Healthy(), implemented in both pkg and ee providers.
User service declares dependency on authz via dependsOn.

* test(integration): use /api/v2/healthz for readiness check, log 503 response body

* fix(factory): replace fmt.Errorf with errors.Newf in tests to satisfy linter

* feat: generate openapi spec

* fix(integration): log errors at error level in healthz readiness check

* test(integration): log and assert healthz response in test_setup

* feat(user): implement ServiceWithHealthy for user service

User service signals healthy after successful root user reconciliation
or immediately when disabled. User Service interface now embeds
factory.ServiceWithHealthy.

* fix(factory): reflect service names as strings

* fix(apiserver): document health 503 responses

* feat: generate openapi spec
2026-03-23 18:46:15 +00:00

83 lines
2.1 KiB
Go

package factory
import "context"
// Service is the core lifecycle interface for long-running services.
type Service interface {
// Starts a service. It should block and should not return until the service is stopped or it fails.
Start(context.Context) error
// Stops a service.
Stop(context.Context) error
}
// Healthy is an optional interface that services can implement to signal
// when they have completed startup and are ready to serve.
// Services that do not implement this interface are considered healthy
// immediately after Start() is called.
type Healthy interface {
// Healthy returns a channel that is closed when the service is healthy.
Healthy() <-chan struct{}
}
// ServiceWithHealthy is a Service that explicitly signals when it is healthy.
type ServiceWithHealthy interface {
Service
Healthy
}
// NamedService is a Service with a Name and optional dependencies.
type NamedService interface {
Named
ServiceWithHealthy
// DependsOn returns the names of services that must be healthy before this service starts.
DependsOn() []Name
}
// closedC is a pre-closed channel returned for services that don't implement Healthy.
var closedC = func() chan struct{} {
c := make(chan struct{})
close(c)
return c
}()
type namedService struct {
name Name
dependsOn []Name
service Service
}
// NewNamedService wraps a Service with a Name and optional dependency names.
func NewNamedService(name Name, service Service, dependsOn ...Name) NamedService {
return &namedService{
name: name,
dependsOn: dependsOn,
service: service,
}
}
func (s *namedService) Name() Name {
return s.name
}
func (s *namedService) DependsOn() []Name {
return s.dependsOn
}
func (s *namedService) Start(ctx context.Context) error {
return s.service.Start(ctx)
}
func (s *namedService) Stop(ctx context.Context) error {
return s.service.Stop(ctx)
}
// Healthy delegates to the underlying service if it implements Healthy,
// otherwise returns an already-closed channel (immediately healthy).
func (s *namedService) Healthy() <-chan struct{} {
if h, ok := s.service.(Healthy); ok {
return h.Healthy()
}
return closedC
}