README
¶
pkg/monitor
pkg/monitor is the polling monitor framework for region.
Unlike handlers and managers, this layer is not driving desired-state reconciliation directly. It periodically inspects live provider-backed resource state and projects that back into:
- Kubernetes status updates
- structured transition logs
- OpenTelemetry metrics
Architecturally, this is important because it is a cross-resource, potentially cross-repo pattern rather than a one-off local helper. It is where the platform admits that some lifecycle truth must be polled and observed, not only pushed by controller watches.
Current Shape
- builds a shared provider registry
- creates OTel instruments
- runs one or more
Checkers on a poll interval - logs and continues on non-fatal per-check failures
Right now the only checker is health/server, but the abstraction is clearly intended to allow additional monitor classes later.
Caveats
- This is polling by design. That makes it simpler and more decoupled, but also means timeliness and overhead are governed by poll period rather than watches.
- Provider/cache readiness is part of monitor startup, but a region whose provider cannot be initialized is skipped rather than taking the process down; see providers. Servers in an unavailable region are absent from that poll's checks and from the state gauge, and keep their last known condition rather than being written to a wrong one, per health/server.
Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
Types ¶
type Checker ¶
type Checker interface {
// Check does whatever the checker is checking for.
Check(ctx context.Context) error
}
Checker is an interface that monitors must implement.
type Options ¶
type Options struct {
// CoreOptions provides --namespace, --otlp-endpoint, logging, and OTel setup.
CoreOptions options.CoreOptions
// contains filtered or unexported fields
}
Options allow modification of parameters via the CLI.
Source Files
Directories