[CELEBORN-1726] Update WorkerInfo when transition worker state

### What changes were proposed in this pull request? Update WorkerInfo when transition worker state ### Why are the changes needed? When we send a `getWorkerInfo` request to the `Worker` node after this worker has became `Decommission` state , it does not return the correct state. note: If we send this request to `Master` node instead of `Worker`, the return value is correct. We should update the worker state also for worker node itself. ### Does this PR introduce _any_ user-facing change? Yes ### How was this patch tested? Mannually Closes #2930 from reswqa/worker_info. Authored-by: Weijie Guo <[email protected]> Signed-off-by: Shuang <[email protected]>
apache · Nov 20, 2024 · bb96a5f · bb96a5f
1 parent 1d0032b
commit bb96a5f
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 2 deletions.
diff --git a/worker/src/main/scala/org/apache/celeborn/service/deploy/worker/WorkerStatusManager.scala b/worker/src/main/scala/org/apache/celeborn/service/deploy/worker/WorkerStatusManager.scala
@@ -129,6 +129,7 @@ private[celeborn] class WorkerStatusManager(conf: CelebornConf) extends Logging
     if (allowStates != null && allowStates.contains(state)) {
       logInfo(s"Worker transition status from ${currentWorkerStatus.getState} to $state.")
       currentWorkerStatus = new WorkerStatus(state.getNumber, System.currentTimeMillis())
+      worker.workerInfo.setWorkerStatus(currentWorkerStatus)
     } else {
       logWarning(
         s"Worker transition status from ${currentWorkerStatus.getState} to $state is not allowed.")

diff --git a/...r/src/test/scala/org/apache/celeborn/service/deploy/worker/WorkerStatusManagerSuite.scala b/...r/src/test/scala/org/apache/celeborn/service/deploy/worker/WorkerStatusManagerSuite.scala
@@ -19,13 +19,17 @@ package org.apache.celeborn.service.deploy.worker
 
 import java.util.concurrent.atomic.AtomicBoolean
 
-import com.google.common.collect.Sets
+import com.google.common.collect.{Maps, Sets}
 import org.junit.Assert
 import org.mockito.MockitoSugar._
 import org.scalatest.funsuite.AnyFunSuite
 
 import org.apache.celeborn.common.CelebornConf
+import org.apache.celeborn.common.identity.UserIdentifier
+import org.apache.celeborn.common.meta.WorkerInfo
 import org.apache.celeborn.common.protocol.{PbWorkerStatus, WorkerEventType}
+import org.apache.celeborn.common.quota.ResourceConsumption
+import org.apache.celeborn.common.util.JavaUtils
 import org.apache.celeborn.service.deploy.worker.storage.StorageManager
 
 class WorkerStatusManagerSuite extends AnyFunSuite {
@@ -36,15 +40,27 @@ class WorkerStatusManagerSuite extends AnyFunSuite {
     worker = mock[Worker]
     val storageManager = mock[StorageManager]
     val shuffleKeys = Sets.newHashSet("test")
+    val workerInfo = new WorkerInfo(
+      "host",
+      0,
+      0,
+      0,
+      0,
+      0,
+      Maps.newHashMap(),
+      JavaUtils.newConcurrentHashMap[UserIdentifier, ResourceConsumption])
     when(storageManager.shuffleKeySet()).thenReturn(shuffleKeys)
     when(worker.storageManager).thenReturn(storageManager)
     when(worker.shutdown).thenReturn(new AtomicBoolean())
-
+    when(worker.workerInfo).thenReturn(workerInfo)
     val statusManager = new WorkerStatusManager(conf)
     statusManager.init(worker)
 
     statusManager.doTransition(WorkerEventType.DecommissionThenIdle)
     Assert.assertEquals(statusManager.getWorkerState(), PbWorkerStatus.State.InDecommissionThenIdle)
+    Assert.assertEquals(
+      worker.workerInfo.getWorkerStatus().getStateValue,
+      PbWorkerStatus.State.InDecommissionThenIdle.getNumber)
 
     // Rerun state Transition
     statusManager.doTransition(WorkerEventType.DecommissionThenIdle)