refactor(scheduler): make bin packing a generic utility per #534 review

apps/optimizer/schedulerapp/src/main/java/com/linkedin/openhouse/optimizer/scheduler/SchedulerApplication.java

@@ -1,7 +1,5 @@
 package com.linkedin.openhouse.optimizer.scheduler;
 
-import com.linkedin.openhouse.optimizer.model.OperationTypeDto;
-import java.util.Map;
 import lombok.extern.slf4j.Slf4j;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.boot.CommandLineRunner;
@@ -26,29 +24,27 @@
 public class SchedulerApplication implements CommandLineRunner, ExitCodeGenerator {
 
   private final SchedulerRunner runner;
-  private final Map<OperationTypeDto, BinPacker> binPackers;
   private int exitCode = 0;
 
   @Autowired
-  public SchedulerApplication(SchedulerRunner runner, Map<OperationTypeDto, BinPacker> binPackers) {
+  public SchedulerApplication(SchedulerRunner runner) {
     this.runner = runner;
-    this.binPackers = binPackers;
   }
 
   public static void main(String[] args) {
     System.exit(SpringApplication.exit(SpringApplication.run(SchedulerApplication.class, args)));
   }
 
   /**
-   * Runs the scheduler once per registered {@link BinPacker} per process invocation. Each call is
-   * scoped to one operation type. Any thrown exception is logged and surfaces as a non-zero exit
-   * code via {@link #getExitCode()} after the context is shut down cleanly.
+   * Runs the scheduler once per registered operation type per process invocation. Any thrown
+   * exception is logged and surfaces as a non-zero exit code via {@link #getExitCode()} after the
+   * context is shut down cleanly.
    */
   @Override
   public void run(String... args) {
     try {
-      log.info("Scheduler starting; operation types: {}", binPackers.keySet());
-      binPackers.keySet().forEach(runner::schedule);
+      log.info("Scheduler starting; operation types: {}", runner.getRegisteredOperationTypes());
+      runner.getRegisteredOperationTypes().forEach(runner::schedule);
       log.info("Scheduler completed successfully");
     } catch (Exception e) {
       log.error("Scheduler failed", e);

apps/optimizer/schedulerapp/src/main/resources/application.properties

@@ -6,6 +6,9 @@ spring.datasource.username=${OPTIMIZER_DB_USER:sa}
 spring.datasource.password=${OPTIMIZER_DB_PASSWORD:}
 spring.jpa.hibernate.ddl-auto=none
 optimizer.scheduler.jobs.base-uri=${JOBS_BASE_URI:http://localhost:8002}
+# Per-bin caps for ORPHAN_FILES_DELETION; 0 disables a dimension. File count is the OFD cost
+# driver — per-file list, manifest joins, and delete calls dominate, independent of file size.
 optimizer.scheduler.ofd.max-files-per-bin=${SCHEDULER_OFD_MAX_FILES_PER_BIN:1000000}
+optimizer.scheduler.ofd.max-tables-per-bin=${SCHEDULER_OFD_MAX_TABLES_PER_BIN:50}
 optimizer.scheduler.results-endpoint=${SCHEDULER_RESULTS_ENDPOINT:http://openhouse-optimizer:8080/v1/optimizer/operations}
 optimizer.scheduler.cluster-id=${SCHEDULER_CLUSTER_ID:LocalHadoopCluster}