Cluster overview page

ui/.env

@@ -1,4 +1,4 @@
-NEXT_PUBLIC_KEYCLOAK_URL="https://console-keycloak.mycluster-us-east-107719-da779aef12eee96bd4161f4e402b70ec-0000.us-east.containers.appdomain.cloud/realms/demo"
+#NEXT_PUBLIC_KEYCLOAK_URL="https://console-keycloak.mycluster-us-east-107719-da779aef12eee96bd4161f4e402b70ec-0000.us-east.containers.appdomain.cloud/realms/demo"
 BACKEND_URL=https://console-api-eyefloaters-dev.mycluster-us-east-107719-da779aef12eee96bd4161f4e402b70ec-0000.us-east.containers.appdomain.cloud
 LOG_LEVEL="info"
-CONSOLE_MODE="read-write"
+CONSOLE_MODE="read-write"

ui/.storybook/preview-head.html

@@ -0,0 +1,5 @@
+<style>
+    html, body, #storybook-root {
+        height: 100%;
+    }
+</style>

ui/.storybook/preview.tsx

@@ -0,0 +1,32 @@
+import NextIntlProvider from "@/app/[locale]/NextIntlProvider";
+import { Page } from "@patternfly/react-core";
+import type { Preview } from "@storybook/react";
+import "../app/globals.css";
+import messages from "../messages/en.json";
+
+const preview: Preview = {
+  parameters: {
+    nextjs: {
+      appDirectory: true,
+    },
+    actions: { argTypesRegex: "^on[A-Z].*" },
+    controls: {
+      matchers: {
+        color: /(background|color)$/i,
+        date: /Date$/i,
+      },
+    },
+    layout: "fullscreen",
+  },
+  decorators: [
+    (Story) => (
+      <NextIntlProvider locale={"en"} messages={messages}>
+        <Page>
+          <Story />
+        </Page>
+      </NextIntlProvider>
+    ),
+  ],
+};
+
+export default preview;

ui/api/kafka/actions.ts

@@ -2,11 +2,25 @@
 import { getHeaders } from "@/api/api";
 import {
   ClusterDetail,
+  ClusterKpis,
+  ClusterKpisSchema,
   ClusterList,
+  ClusterMetricRange,
+  ClusterMetricRangeSchema,
   ClusterResponse,
   ClustersResponseSchema,
 } from "@/api/kafka/schema";
 import { logger } from "@/utils/logger";
+import groupBy from "lodash.groupby";
+import { PrometheusDriver } from "prometheus-query";
+import * as ranges from "./ranges.promql";
+import { values } from "./values.promql";
+
+export type Range = keyof typeof ranges;
+
+const prom = new PrometheusDriver({
+  endpoint: process.env.CONSOLE_METRICS_PROMETHEUS_URL,
+});
 
 const log = logger.child({ module: "kafka-api" });
 
@@ -27,15 +41,191 @@ export async function getKafkaClusters(): Promise<ClusterList[]> {
 export async function getKafkaCluster(
   clusterId: string,
 ): Promise<ClusterDetail | null> {
-  const url = `${process.env.BACKEND_URL}/api/kafkas/${clusterId}/?fields%5Bkafkas%5D=name,namespace,creationTimestamp,nodes,controller,authorizedOperations,bootstrapServers,authType,metrics`;
+  const url = `${process.env.BACKEND_URL}/api/kafkas/${clusterId}/?fields%5Bkafkas%5D=name,namespace,creationTimestamp,status,kafkaVersion,nodes,controller,authorizedOperations,bootstrapServers,authType`;
   try {
     const res = await fetch(url, {
       headers: await getHeaders(),
+      cache: "force-cache",
     });
     const rawData = await res.json();
+    log.debug(rawData, "getKafkaCluster response");
     return ClusterResponse.parse(rawData).data;
   } catch (err) {
-    log.error(err, "getKafkaCluster");
+    log.error({ err, clusterId }, "getKafkaCluster");
+    return null;
+  }
+}
+
+export async function getKafkaClusterKpis(
+  clusterId: string,
+): Promise<{ cluster: ClusterDetail; kpis: ClusterKpis } | null> {
+  try {
+    const cluster = await getKafkaCluster(clusterId);
+    if (!cluster) {
+      return null;
+    }
+
+    const valuesRes = await prom.instantQuery(
+      values(
+        cluster.attributes.namespace,
+        cluster.attributes.name,
+        cluster.attributes.controller.id,
+      ),
+    );
+
+    /*
+    Prometheus returns the data unaggregated. Eg.
+
+    [
+      {
+        "metric": {
+          "labels": {
+            "__console_metric_name__": "broker_state",
+            "nodeId": "2"
+          }
+        },
+        "value": {
+          "time": "2023-12-12T16:00:53.381Z",
+          "value": 3
+         }
+      },
+      ...
+    ]
+
+    We start by flattening the labels, and then group by metric name
+     */
+    const groupedMetrics = groupBy(
+      valuesRes.result.map((serie) => ({
+        metric: serie.metric.labels.__console_metric_name__,
+        nodeId: serie.metric.labels.nodeId,
+        time: serie.value.time,
+        value: serie.value.value,
+      })),
+      (v) => v.metric,
+    );
+
+    /*
+    Now we want to transform the data in something easier to work with in the UI.
+
+    Some are totals, in an array form with a single entry; we just need the number. These will look like a metric:value
+    mapping.
+
+    Some KPIs are provided split by broker id. Of these, some are counts (identified by the string `_count` in the
+    metric name), and some are other infos. Both will be grouped by nodeId.
+    The `_count` metrics will have a value with two properties, `byNode` and `total`. `byNode` will hold the grouping. `total` will
+    have the sum of all the counts.
+    Other metrics will look like a metric:[node:value] mapping.
+
+    Expected result:
+    {
+      "broker_state": {
+        "0": 3,
+        "1": 3,
+        "2": 3
+      },
+      "total_topics": 5,
+      "total_partitions": 55,
+      "underreplicated_topics": 0,
+      "replica_count": {
+        "byNode": {
+          "0": 57,
+          "1": 54,
+          "2": 54
+        },
+        "total": 165
+      },
+      "leader_count": {
+        "byNode": {
+          "0": 19,
+          "1": 18,
+          "2": 18
+        },
+        "total": 55
+      }
+    }
+     */
+    const kpis = Object.fromEntries(
+      Object.entries(groupedMetrics).map(([metric, value]) => {
+        const total = value.reduce((acc, v) => acc + v.value, 0);
+        if (value.find((v) => v.nodeId)) {
+          const byNode = Object.fromEntries(
+            value.map(({ nodeId, value }) =>
+              nodeId ? [nodeId, value] : ["value", value],
+            ),
+          );
+          return metric.includes("_count")
+            ? [
+                metric,
+                {
+                  byNode,
+                  total,
+                },
+              ]
+            : [metric, byNode];
+        } else {
+          return [metric, total];
+        }
+      }),
+    );
+    log.debug({ kpis, clusterId }, "getKafkaClusterKpis");
+    return {
+      cluster,
+      kpis: ClusterKpisSchema.parse(kpis),
+    };
+  } catch (err) {
+    log.error({ err, clusterId }, "getKafkaClusterKpis");
+    return null;
+  }
+}
+
+export async function getKafkaClusterMetrics(
+  clusterId: string,
+  metrics: Array<Range>,
+): Promise<{
+  cluster: ClusterDetail;
+  ranges: Record<Range, ClusterMetricRange>;
+} | null> {
+  async function getRange(namespace: string, name: string, metric: Range) {
+    const start = new Date().getTime() - 1 * 60 * 60 * 1000;
+    const end = new Date();
+    const step = 60 * 2;
+    const seriesRes = await prom.rangeQuery(
+      ranges[metric](namespace, name),
+      start,
+      end,
+      step,
+    );
+    const range = Object.fromEntries(
+      seriesRes.result.flatMap((serie) =>
+        serie.values.map((v: any) => [new Date(v.time).getTime(), v.value]),
+      ),
+    );
+    return [metric, ClusterMetricRangeSchema.parse(range)];
+  }
+
+  try {
+    const cluster = await getKafkaCluster(clusterId);
+    if (!cluster) {
+      return null;
+    }
+
+    const rangesRes = Object.fromEntries(
+      await Promise.all(
+        metrics.map((m) =>
+          getRange(cluster.attributes.namespace, cluster.attributes.name, m),
+        ),
+      ),
+    );
+    log.debug(
+      { ranges: rangesRes, clusterId, metric: metrics },
+      "getKafkaClusterMetric",
+    );
+    return {
+      cluster,
+      ranges: rangesRes,
+    };
+  } catch (err) {
+    log.error({ err, clusterId, metric: metrics }, "getKafkaClusterMetric");
     return null;
   }
 }

ui/api/kafka/ranges.promql.ts

@@ -0,0 +1,95 @@
+export const cpu = (namespace: string, cluster: string) => `
+  sum by (nodeId, __console_metric_name__) (
+    label_replace(
+      label_replace(
+        rate(container_cpu_usage_seconds_total{namespace="${namespace}",pod=~"${cluster}-.+-\\\\d+"}[1m]),
+        "nodeId",
+        "$1",
+        "pod",
+        ".+-(\\\\d+)"
+      ),
+      "__console_metric_name__",
+      "cpu_usage_seconds",
+      "",
+      ""
+    )
+  )
+`;
+
+export const memory = (namespace: string, cluster: string) => `
+  sum by (nodeId, __console_metric_name__) (
+    label_replace(
+      label_replace(
+        container_memory_usage_bytes{namespace="${namespace}",pod=~"${cluster}-.+-\\\\d+"},
+        "nodeId",
+        "$1",
+        "pod",
+        ".+-(\\\\d+)"
+      ),
+      "__console_metric_name__",
+      "memory_usage_bytes",
+      "",
+      ""
+    )
+  )
+`;
+
+export const incomingByteRate = (namespace: string, cluster: string) => `
+  sum by (__console_metric_name__) (
+    label_replace(
+      irate(kafka_server_brokertopicmetrics_bytesin_total{topic!="",namespace="${namespace}",pod=~"${cluster}-.+-\\\\d+",strimzi_io_kind="Kafka"}[5m]),
+      "__console_metric_name__",
+      "incoming_byte_rate",
+      "",
+      ""
+    )
+  )
+`;
+
+export const outgoingByteRate = (namespace: string, cluster: string) => `
+  sum by (__console_metric_name__) (
+    label_replace(
+      irate(kafka_server_brokertopicmetrics_bytesout_total{topic!="",namespace="${namespace}",pod=~"${cluster}-.+-\\\\d+",strimzi_io_kind="Kafka"}[5m]),
+      "__console_metric_name__",
+      "outgoing_byte_rate",
+      "",
+      ""
+    )
+  )
+`;
+
+export const volumeCapacity = (namespace: string, cluster: string) => `
+  sum by (nodeId, __console_metric_name__) (
+    label_replace(
+      label_replace(
+        kubelet_volume_stats_capacity_bytes{namespace="${namespace}",persistentvolumeclaim=~"data(?:-\\\\d+)?-${cluster}-.+-\\\\d+"},
+        "nodeId",
+        "$1",
+        "persistentvolumeclaim",
+        ".+-(\\\\d+)"
+      ),
+      "__console_metric_name__",
+      "volume_stats_capacity_bytes",
+      "",
+      ""
+    )
+  )
+`;
+
+export const volumeUsed = (namespace: string, cluster: string) => `
+  sum by (nodeId, __console_metric_name__) (
+    label_replace(
+      label_replace(
+        kubelet_volume_stats_used_bytes{namespace="${namespace}",persistentvolumeclaim=~"data(?:-\\\\d+)?-${cluster}-.+-\\\\d+"},
+        "nodeId",
+        "$1",
+        "persistentvolumeclaim",
+        ".+-(\\\\d+)"
+      ),
+      "__console_metric_name__",
+      "volume_stats_used_bytes",
+      "",
+      ""
+    )
+  )
+`;