[Feature] Performance improvements and add airforce, army, navy, NLM agencies -- Production

.github/workflows/ci.yml

@@ -81,7 +81,6 @@ jobs:
       CF_ORGANIZATION_NAME: ${{ vars.CF_ORGANIZATION_NAME }}
       CF_SPACE_NAME: ${{ vars.CF_SPACE_NAME_DEV }}
       DB_SERVICE_NAME: ${{ vars.DB_SERVICE_NAME_DEV }}
-      MESSAGE_QUEUE_DATABASE_NAME: ${{ vars.MESSAGE_QUEUE_DATABASE_NAME }}
       MESSAGE_QUEUE_NAME: ${{ vars.MESSAGE_QUEUE_NAME }}
       NEW_RELIC_APP_NAME: ${{ vars.NEW_RELIC_APP_NAME_DEV }}
       PROXY_FQDN: ${{ vars.PROXY_FQDN_DEV }}
@@ -111,7 +110,6 @@ jobs:
       CF_ORGANIZATION_NAME: ${{ vars.CF_ORGANIZATION_NAME }}
       CF_SPACE_NAME: ${{ vars.CF_SPACE_NAME_STG }}
       DB_SERVICE_NAME: ${{ vars.DB_SERVICE_NAME_STG }}
-      MESSAGE_QUEUE_DATABASE_NAME: ${{ vars.MESSAGE_QUEUE_DATABASE_NAME }}
       MESSAGE_QUEUE_NAME: ${{ vars.MESSAGE_QUEUE_NAME }}
       NEW_RELIC_APP_NAME: ${{ vars.NEW_RELIC_APP_NAME_STG }}
       PROXY_FQDN: ${{ vars.PROXY_FQDN_STG }}
@@ -141,7 +139,6 @@ jobs:
       CF_ORGANIZATION_NAME: ${{ vars.CF_ORGANIZATION_NAME }}
       CF_SPACE_NAME: ${{ vars.CF_SPACE_NAME_PRD }}
       DB_SERVICE_NAME: ${{ vars.DB_SERVICE_NAME_PRD }}
-      MESSAGE_QUEUE_DATABASE_NAME: ${{ vars.MESSAGE_QUEUE_DATABASE_NAME }}
       MESSAGE_QUEUE_NAME: ${{ vars.MESSAGE_QUEUE_NAME }}
       NEW_RELIC_APP_NAME: ${{ vars.NEW_RELIC_APP_NAME_PRD }}
       PROXY_FQDN: ${{ vars.PROXY_FQDN_PRD }}

.github/workflows/deploy.yml

@@ -27,9 +27,6 @@ on:
       DB_SERVICE_NAME:
         required: true
         type: string
-      MESSAGE_QUEUE_DATABASE_NAME:
-        required: true
-        type: string
       MESSAGE_QUEUE_NAME:
         required: true
         type: string
@@ -72,7 +69,6 @@ env:
   CF_SPACE_NAME: ${{ inputs.CF_SPACE_NAME }}
   DB_SERVICE_NAME: ${{ inputs.DB_SERVICE_NAME }}
   GA4_CREDS: ${{ secrets.GA4_CREDS }}
-  MESSAGE_QUEUE_DATABASE_NAME: ${{ inputs.MESSAGE_QUEUE_DATABASE_NAME }}
   MESSAGE_QUEUE_NAME: ${{ inputs.MESSAGE_QUEUE_NAME }}
   NEW_RELIC_APP_NAME: ${{ inputs.NEW_RELIC_APP_NAME }}
   NEW_RELIC_LICENSE_KEY: ${{ secrets.NEW_RELIC_LICENSE_KEY }}

.github/workflows/manual_deploy_to_dev.yml

@@ -15,7 +15,6 @@ jobs:
       CF_ORGANIZATION_NAME: ${{ vars.CF_ORGANIZATION_NAME }}
       CF_SPACE_NAME: ${{ vars.CF_SPACE_NAME_DEV }}
       DB_SERVICE_NAME: ${{ vars.DB_SERVICE_NAME_DEV }}
-      MESSAGE_QUEUE_DATABASE_NAME: ${{ vars.MESSAGE_QUEUE_DATABASE_NAME }}
       MESSAGE_QUEUE_NAME: ${{ vars.MESSAGE_QUEUE_NAME }}
       NEW_RELIC_APP_NAME: ${{ vars.NEW_RELIC_APP_NAME_DEV }}
       PROXY_FQDN: ${{ vars.PROXY_FQDN_DEV }}

deploy/agencies.json

@@ -24,11 +24,21 @@
     "agencyName": "agriculture",
     "awsBucketPath": "data/agriculture"
   },
+  {
+    "analyticsReportIds": "470235781",
+    "agencyName": "air-force",
+    "awsBucketPath": "data/air-force"
+  },
   {
     "analyticsReportIds": "395213963",
     "agencyName": "american-battle-monuments-commission",
     "awsBucketPath": "data/american-battle-monuments-commission"
   },
+  {
+    "analyticsReportIds": "470089273",
+    "agencyName": "army",
+    "awsBucketPath": "data/army"
+  },
   {
     "analyticsReportIds": "395253935",
     "agencyName": "commerce",
@@ -264,6 +274,11 @@
     "agencyName": "national-labor-relations-board",
     "awsBucketPath": "data/national-labor-relations-board"
   },
+  {
+    "analyticsReportIds": "470944610",
+    "agencyName": "national-library-medicine",
+    "awsBucketPath": "data/national-library-medicine"
+  },
   {
     "analyticsReportIds": "425930242",
     "agencyName": "national-mediation-board",
@@ -294,6 +309,11 @@
     "agencyName": "national-transportation-safety-board",
     "awsBucketPath": "data/national-transportation-safety-board"
   },
+  {
+    "analyticsReportIds": "470101393",
+    "agencyName": "navy",
+    "awsBucketPath": "data/navy"
+  },
   {
     "analyticsReportIds": "395460734",
     "agencyName": "nuclear-regulatory-commission",

deploy/api.sh

@@ -3,4 +3,4 @@
 export ANALYTICS_REPORTS_PATH=reports/api.json
 export ANALYTICS_SCRIPT_NAME=api.sh
 
-$ANALYTICS_ROOT_PATH/bin/analytics-publisher --debug --write-to-database --output /tmp --agenciesFile=$ANALYTICS_ROOT_PATH/deploy/agencies.json
+$ANALYTICS_ROOT_PATH/bin/analytics-publisher --debug --write-to-database --agenciesFile=$ANALYTICS_ROOT_PATH/deploy/agencies.json

deploy/cron.js

@@ -61,18 +61,18 @@ const daily_run = () => {
   runScriptWithLogName(`${scriptRootPath}/daily.sh`, "daily.sh");
 };
 
-const hourly_run = () => {
+/*const hourly_run = () => {
   runScriptWithLogName(`${scriptRootPath}/hourly.sh`, "hourly.sh");
-};
+};*/
 
 const realtime_run = () => {
   runScriptWithLogName(`${scriptRootPath}/realtime.sh`, "realtime.sh");
 };
 
 /**
-  Daily reports run every morning at 10 AM UTC.
-  This calculates the offset between now and then for the next scheduled run.
-*/
+ * Daily and API reports run every morning at 10 AM UTC.
+ * This calculates the offset between now and then for the next scheduled run.
+ */
 const calculateNextDailyRunTimeOffset = () => {
   const currentTime = new Date();
   const nextRunTime = new Date(
@@ -85,26 +85,36 @@ const calculateNextDailyRunTimeOffset = () => {
 };
 
 /**
- * All scripts run immediately upon application start (with a 10 second delay
+ * All scripts run immediately upon application start (with a 60 second delay
  * between each so that they don't all run at once), then run again at intervals
  * going forward.
  */
 setTimeout(realtime_run, 1000 * 10);
-setTimeout(hourly_run, 1000 * 20);
-setTimeout(daily_run, 1000 * 30);
-setTimeout(api_run, 1000 * 40);
+// setTimeout(hourly_run, 1000 * 70); No hourly reports exist at this time.
+setTimeout(daily_run, 1000 * 70);
+setTimeout(api_run, 1000 * 130);
 
-// daily
+// Daily and API recurring script run setup.
 // Runs at 10 AM UTC, then every 24 hours afterwards
 setTimeout(() => {
-  daily_run();
-  setInterval(daily_run, 1000 * 60 * 60 * 24);
-  // API
-  api_run();
-  setInterval(api_run, 1000 * 60 * 60 * 24);
+  // Offset the daily script run by 30 seconds so that it never runs in parallel
+  // with the realtime script in order to save memory/CPU.
+  setTimeout(() => {
+    daily_run();
+    setInterval(daily_run, 1000 * 60 * 60 * 24);
+  }, 1000 * 30);
+
+  // setTimeout(hourly_run, 1000 * 60);
+
+  // Offset the API script run by 90 seconds so that it never runs in parallel
+  // with the daily or realtime scripts in order to save memory/CPU.
+  setTimeout(() => {
+    api_run();
+    setInterval(api_run, 1000 * 60 * 60 * 24);
+  }, 1000 * 90);
 }, calculateNextDailyRunTimeOffset());
-// hourly
-setInterval(hourly_run, 1000 * 60 * 60);
-// realtime. Runs every 15 minutes.
-// Google updates realtime reports every 30 minutes, so there is some overlap.
+// hourly (no hourly reports exist at this time).
+// setInterval(hourly_run, 1000 * 60 * 60);
+
+// Realtime recurring script run setup. Runs every 15 minutes.
 setInterval(realtime_run, 1000 * 60 * 15);

index.js

@@ -1,10 +1,12 @@
 const { AsyncLocalStorage } = require("node:async_hooks");
+const knex = require("knex");
 const PgBoss = require("pg-boss");
 const util = require("util");
 const AppConfig = require("./src/app_config");
 const ReportProcessingContext = require("./src/report_processing_context");
 const Logger = require("./src/logger");
 const Processor = require("./src/processor");
+const PgBossKnexAdapter = require("./src/pg_boss_knex_adapter");
 
 /**
  * Gets an array of JSON report objects from the application confing, then runs
@@ -80,7 +82,7 @@ async function _processReport(appConfig, context, reportConfig, processor) {
       await processor.processChain(context);
       logger.info("Processing complete");
     } catch (e) {
-      logger.error("Encountered an error");
+      logger.error("Encountered an error during report processing");
       logger.error(util.inspect(e));
     }
   });
@@ -121,8 +123,8 @@ async function runQueuePublish(options = {}) {
     agencyName: appConfig.agencyLogName,
     scriptName: appConfig.scriptName,
   });
-  const queueClient = await _initQueueClient(appConfig, appLogger);
-  const queue = "analytics-reporter-job-queue";
+  const knexInstance = await knex(appConfig.knexConfig);
+  const queueClient = await _initQueueClient(knexInstance, appLogger);
 
   for (const agency of agencies) {
     for (const reportConfig of reportConfigs) {
@@ -134,7 +136,7 @@ async function runQueuePublish(options = {}) {
       });
       try {
         let jobId = await queueClient.send(
-          queue,
+          appConfig.messageQueueName,
           _createQueueMessage(
             options,
             agency,
@@ -151,13 +153,17 @@ async function runQueuePublish(options = {}) {
         );
         if (jobId) {
           reportLogger.info(
-            `Created job in queue: ${queue} with job ID: ${jobId}`,
+            `Created job in queue: ${appConfig.messageQueueName} with job ID: ${jobId}`,
           );
         } else {
-          reportLogger.info(`Found a duplicate job in queue: ${queue}`);
+          reportLogger.info(
+            `Found a duplicate job in queue: ${appConfig.messageQueueName}`,
+          );
         }
       } catch (e) {
-        reportLogger.error(`Error sending to queue: ${queue}`);
+        reportLogger.error(
+          `Error sending to queue: ${appConfig.messageQueueName}`,
+        );
         reportLogger.error(util.inspect(e));
       }
     }
@@ -169,6 +175,9 @@ async function runQueuePublish(options = {}) {
   } catch (e) {
     appLogger.error("Error stopping queue client");
     appLogger.error(util.inspect(e));
+  } finally {
+    appLogger.debug(`Destroying database connection pool`);
+    knexInstance.destroy();
   }
 }
 
@@ -189,10 +198,10 @@ function _initAgencies(agencies_file) {
   return Array.isArray(agencies) ? agencies : legacyAgencies;
 }
 
-async function _initQueueClient(appConfig, logger) {
+async function _initQueueClient(knexInstance, logger) {
   let queueClient;
   try {
-    queueClient = new PgBoss(appConfig.messageQueueDatabaseConnection);
+    queueClient = new PgBoss({ db: new PgBossKnexAdapter(knexInstance) });
     await queueClient.start();
     logger.debug("Starting queue client");
   } catch (e) {
@@ -230,15 +239,19 @@ function _messagePriority(reportConfig) {
 async function runQueueConsume() {
   const appConfig = new AppConfig();
   const appLogger = Logger.initialize();
-  const queueClient = await _initQueueClient(appConfig, appLogger);
-  const queue = "analytics-reporter-job-queue";
+  const knexInstance = await knex(appConfig.knexConfig);
+  const queueClient = await _initQueueClient(knexInstance, appLogger);
 
   try {
     const context = new ReportProcessingContext(new AsyncLocalStorage());
-    const processor = Processor.buildAnalyticsProcessor(appConfig, appLogger);
+    const processor = Processor.buildAnalyticsProcessor(
+      appConfig,
+      appLogger,
+      knexInstance,
+    );
 
     await queueClient.work(
-      queue,
+      appConfig.messageQueueName,
       { newJobCheckIntervalSeconds: 1 },
       async (message) => {
         appLogger.info("Queue message received");

knexfile.js

@@ -8,6 +8,10 @@ module.exports = {
       password: process.env.POSTGRES_PASSWORD || "123abc",
       port: 5432,
     },
+    pool: {
+      min: 2,
+      max: 10,
+    },
   },
   test: {
     client: "postgresql",
@@ -18,6 +22,10 @@ module.exports = {
       password: process.env.POSTGRES_PASSWORD || "123abc",
       port: 5432,
     },
+    pool: {
+      min: 2,
+      max: 10,
+    },
     migrations: {
       tableName: "knex_migrations",
     },
@@ -31,5 +39,9 @@ module.exports = {
       password: process.env.POSTGRES_PASSWORD,
       ssl: true,
     },
+    pool: {
+      min: 2,
+      max: 10,
+    },
   },
 };

manifest.consumer.yml

@@ -21,7 +21,6 @@ applications:
     ANALYTICS_REPORT_EMAIL: ${ANALYTICS_REPORT_EMAIL}
     AWS_CACHE_TIME: '0'
     GOOGLE_APPLICATION_CREDENTIALS: /home/vcap/app/${ANALYTICS_KEY_FILE_NAME}
-    MESSAGE_QUEUE_DATABASE_NAME: ${MESSAGE_QUEUE_DATABASE_NAME}
     MESSAGE_QUEUE_NAME: ${MESSAGE_QUEUE_NAME}
     NEW_RELIC_APP_NAME: ${NEW_RELIC_APP_NAME}
     NEW_RELIC_LICENSE_KEY: ${NEW_RELIC_LICENSE_KEY}

manifest.publisher.yml

@@ -21,7 +21,6 @@ applications:
     # The default path for reports (used for gov-wide reports)
     AWS_BUCKET_PATH: data/live
     AWS_CACHE_TIME: '0'
-    MESSAGE_QUEUE_DATABASE_NAME: ${MESSAGE_QUEUE_DATABASE_NAME}
     MESSAGE_QUEUE_NAME: ${MESSAGE_QUEUE_NAME}
     NEW_RELIC_APP_NAME: ${NEW_RELIC_APP_NAME}
     NEW_RELIC_LICENSE_KEY: ${NEW_RELIC_LICENSE_KEY}