Capture githubRateLimitRemaining metric in CloudWatch

Author: mikebroberts

src/app/domain/github/crawler/crawlInstallation.ts

@@ -15,7 +15,7 @@ export async function crawlInstallation(
   logger.info(`Crawling Installation for ${installation.accountLogin}`)
   const githubInstallationClient = appState.githubClient.clientForInstallation(installation.installationId)
   await crawlUsers(appState, installation, githubInstallationClient)
-  publishGithubInstallationClientMetrics(installation, githubInstallationClient)
+  publishGithubInstallationClientMetrics(githubInstallationClient)
   const repos = await crawlRepositories(appState, installation, githubInstallationClient)
   // Eventually consider doing some parallelization here (or move back to step function) but
   // need to be careful since GitHub gets twitchy about concurrent requests to the API
@@ -25,6 +25,6 @@ export async function crawlInstallation(
     await crawlPushes(appState, installation, repo, githubInstallationClient)
     await crawlWorkflowRunEvents(appState, installation, repo, lookbackDays, githubInstallationClient)
   }
-  publishGithubInstallationClientMetrics(installation, githubInstallationClient)
+  publishGithubInstallationClientMetrics(githubInstallationClient)
   logger.info('Github Metadata after crawls', { ...githubInstallationClient.meta() })
 }

src/app/outboundInterfaces/githubInstallationClient.ts

@@ -4,7 +4,6 @@ import { RawGithubWorkflowRunEvent } from '../domain/types/rawGithub/RawGithubWo
 import { RawGithubRepository } from '../domain/types/rawGithub/RawGithubRepository'
 import { RawGithubUser } from '../domain/types/rawGithub/RawGithubUser'
 import { RawGithubEvent } from '../domain/types/rawGithub/RawGithubEvent'
-import { GithubInstallation } from '../domain/types/GithubInstallation'
 import { metrics } from '../util/metrics'
 import { MetricUnit } from '@aws-lambda-powertools/metrics'
 
@@ -132,13 +131,13 @@ export type OctokitResponseHeaders = {
 }
 
 // ToEventually - move this into the actual GithubInstallationClient object
-export function publishGithubInstallationClientMetrics(
-  installation: GithubInstallation,
-  githubInstallationClient: GithubInstallationClient
-) {
+export function publishGithubInstallationClientMetrics(githubInstallationClient: GithubInstallationClient) {
+  // Eventually considering adding "installation" as a dimension here to allow different metrics / alarms
+  // for different installations. The reason I didn't just do that immediately is that, for now, Alarms
+  // are defined at deployment time, but installations are a runtime concept.
   const rateLimitMetric = metrics.singleMetric()
-  rateLimitMetric.addDimension('installationAccount', installation.accountLogin)
   rateLimitMetric.addMetric(
+    // ToEventually - this is a shared string with CDK so move to constant
     'githubRateLimitRemaining',
     MetricUnit.Count,
     githubInstallationClient.meta().ratelimitRemaining

src/cdk/stacks/main/MainStack.ts

@@ -9,6 +9,7 @@ import { saveInSSMViaCloudFormation } from '../../support/ssm'
 import { SSM_PARAM_NAMES, SsmParamName } from '../../../multipleContexts/ssmParams'
 import { defineGithubCrawlers } from './githubCrawlers'
 import { ReportingStack } from './reporting/ReportingStack'
+import { defineMonitoring } from './monitoring'
 
 export class MainStack extends Stack {
   constructor(scope: Construct, id: string, props: AllStacksProps) {
@@ -32,6 +33,8 @@ export class MainStack extends Stack {
 
     savePreGeneratedConfiguration(this, props)
 
+    defineMonitoring(this, mainStackProps)
+
     new ReportingStack(this, 'ReportingStack', {
       ...props,
       stackName: `${props.appName}-reporting`

src/cdk/stacks/main/monitoring.ts

@@ -0,0 +1,29 @@
+import { Alarm, ComparisonOperator, Metric, TreatMissingData } from 'aws-cdk-lib/aws-cloudwatch'
+import { Construct } from 'constructs'
+import { MainStackProps } from './mainStackProps'
+import { Duration } from 'aws-cdk-lib'
+import { SnsAction } from 'aws-cdk-lib/aws-cloudwatch-actions'
+import { Topic } from 'aws-cdk-lib/aws-sns'
+
+export function defineMonitoring(scope: Construct, props: MainStackProps) {
+  const githubRateLimitRemainingAlarm = new Alarm(scope, 'Errors', {
+    comparisonOperator: ComparisonOperator.LESS_THAN_THRESHOLD,
+    threshold: 1000,
+    evaluationPeriods: 1,
+    treatMissingData: TreatMissingData.IGNORE,
+    alarmName: `${props.appName} - Github Rate Limit Remaining`,
+    metric: new Metric({
+      namespace: 'cicada',
+      dimensionsMap: {
+        service: props.appName
+      },
+      // ToEventually - this is a shared string with app code so move to constant
+      metricName: 'githubRateLimitRemaining',
+      statistic: 'Minimum',
+      period: Duration.days(1)
+    })
+  })
+
+  const actionTopic = new Topic(scope, 'alarmActionTopic', { topicName: `${props.appName}-alarms` })
+  githubRateLimitRemainingAlarm.addAlarmAction(new SnsAction(actionTopic))
+}