Documentation
¶
Overview ¶
Package metrics provides a UI-independent background collector that samples the active cluster's message counts and broker stats on an interval, derives message-in rates from the count deltas between cycles, and caches the results plus a rolling time-series history so the metrics page can read them without blocking. It mirrors the design of pkg/cluster/collector.go.
Collection is driven purely through the api.KafkaDataSource interface (GetTopics, GetTopicMessageCounts, GetBrokerStats), which operate on the active context — so the collector samples whichever cluster is currently selected. Byte rates require a configured metrics endpoint; that path is a documented stub and byte rates are otherwise api.RateUnknown.
Index ¶
- Constants
- func NewExpositionHandler(src SnapshotLister, enabledFor func(cluster string) bool) http.Handler
- type CollectTickMsg
- type Collector
- func (c *Collector) Active() (api.ClusterMetrics, bool)
- func (c *Collector) ActiveMessagesInHistory() api.TimeSeries
- func (c *Collector) CollectAll(ctx context.Context)
- func (c *Collector) CollectCmd() tea.Cmd
- func (c *Collector) Interval() time.Duration
- func (c *Collector) List() []api.ClusterMetrics
- func (c *Collector) MessagesInHistory(cluster string) api.TimeSeries
- func (c *Collector) SetSettingsResolver(fn func(cluster string) appconfig.MetricsSettings)
- func (c *Collector) Snapshot(cluster string) (api.ClusterMetrics, bool)
- func (c *Collector) TickCmd() tea.Cmd
- type MetricsUpdatedMsg
- type SnapshotLister
Constants ¶
const ExpositionContentType = "text/plain; version=0.0.4; charset=utf-8"
ExpositionContentType is the Prometheus text exposition format content type.
Variables ¶
This section is empty.
Functions ¶
func NewExpositionHandler ¶
func NewExpositionHandler(src SnapshotLister, enabledFor func(cluster string) bool) http.Handler
NewExpositionHandler builds the flag-gated Prometheus exposition handler (MM-16). enabledFor reports whether a cluster opts in to exposition; clusters for which it returns false are 404 on their own path and excluded from the merged global endpoint.
- GET /metrics merged families across all opted-in clusters
- GET /metrics/{cluster} a single cluster (404 unknown or opted-out)
Types ¶
type CollectTickMsg ¶
type CollectTickMsg struct{}
CollectTickMsg triggers a periodic collection; the subscriber turns it into a CollectCmd (see pkg/ui wiring), mirroring the cluster collector.
type Collector ¶
type Collector struct {
// contains filtered or unexported fields
}
Collector samples the active cluster and caches metrics + history. Safe for concurrent use: the background loop writes while the UI reads.
func New ¶
func New(ds api.KafkaDataSource, interval time.Duration, endpointFor func(string) string) *Collector
New builds a Collector. interval <= 0 falls back to the config default. endpointFor may be nil (no metrics endpoints configured anywhere).
func (*Collector) Active ¶
func (c *Collector) Active() (api.ClusterMetrics, bool)
Active returns the cached metrics for the currently-selected cluster.
func (*Collector) ActiveMessagesInHistory ¶
func (c *Collector) ActiveMessagesInHistory() api.TimeSeries
ActiveMessagesInHistory returns the message-in-rate series for the active cluster.
func (*Collector) CollectAll ¶
CollectAll runs one collection cycle against the active cluster and updates the cache atomically. Fetch failures leave the previous snapshot in place.
func (*Collector) CollectCmd ¶
CollectCmd runs a full collection cycle and reports the refreshed active snapshot. While a cycle is already running it does nothing and reports nothing: the running cycle's MetricsUpdatedMsg serves both callers, and one slow cycle can never stack another on top of it against the same brokers.
func (*Collector) List ¶
func (c *Collector) List() []api.ClusterMetrics
List returns cached metrics for all seen clusters in stable order.
func (*Collector) MessagesInHistory ¶
func (c *Collector) MessagesInHistory(cluster string) api.TimeSeries
MessagesInHistory returns the retained message-in-rate series for a cluster.
func (*Collector) SetSettingsResolver ¶
func (c *Collector) SetSettingsResolver(fn func(cluster string) appconfig.MetricsSettings)
SetSettingsResolver installs a resolver for per-cluster metrics settings so the collector can honor Type "JMX" via the Jolokia bridge (MM-17). It is wired from the UI boot path; without it collection stays Prometheus-only.
type MetricsUpdatedMsg ¶
type MetricsUpdatedMsg struct {
Active api.ClusterMetrics
}
MetricsUpdatedMsg is emitted after a collection cycle so the metrics page can re-render from the collector cache without polling.
type SnapshotLister ¶
type SnapshotLister interface {
List() []api.ClusterMetrics
}
SnapshotLister provides the cached per-cluster snapshots the exposition endpoint renders. *Collector satisfies it.
Source Files
Directories
¶
| Path | Synopsis |
|---|---|
|
Package graphs is a built-in catalog of Prometheus graph descriptions and the logic that renders their query templates and executes them against the optional time-series backend (pkg/metrics/promquery).
|
Package graphs is a built-in catalog of Prometheus graph descriptions and the logic that renders their query templates and executes them against the optional time-series backend (pkg/metrics/promquery). |
|
Package jolokia is an HTTP-bridge collector for JMX metrics exposed through a Jolokia agent.
|
Package jolokia is an HTTP-bridge collector for JMX metrics exposed through a Jolokia agent. |
|
Package promquery is a minimal client for the Prometheus HTTP query API (/api/v1/query and /api/v1/query_range) built on the standard library only.
|
Package promquery is a minimal client for the Prometheus HTTP query API (/api/v1/query and /api/v1/query_range) built on the standard library only. |