temporarily disable clearing on-going, so that we don't trigger recon…

…struction repeatitively
jiajunmao · Aug 22, 2024 · 4426deb · 4426deb
1 parent 7346580
commit 4426deb
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 1 deletion.
diff --git a/...ject/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BPServiceActor.java b/...ject/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/BPServiceActor.java
@@ -561,8 +561,12 @@ HeartbeatResponse sendHeartBeat(boolean requestBlockReportLease)
     // Loop around all the volumes on this datanode, and check for ZFS failures
     for (DatanodeVolumeInfo volInfo : dn.getVolumeReport()) {
       if (volInfo.getStorageType() == StorageType.ZFS) {
+        LOG.info("Heartbeating ZFS");
         if (ErasureCodingWorker.ongoingRepairs.isEmpty()) {
           List<DnodeAttributes> dnodes = new Tools().getFailedChunks("pool");
+          if (dnodes == null) {
+            dnodes = new ArrayList<>();
+          }
           // This means that there is ZFS local failure
           for (DnodeAttributes dnode : dnodes) {
             // Enable this during real testing

diff --git a/...java/org/apache/hadoop/hdfs/server/datanode/erasurecode/StripedZfsBlockReconstructor.java b/...java/org/apache/hadoop/hdfs/server/datanode/erasurecode/StripedZfsBlockReconstructor.java
@@ -97,7 +97,13 @@ public void run() {
 
       // MLEC - clear repair bookkeeping maps
       LOG.info("Clearing ongoing repair map entry {}", getBlockGroup().getBlockId());
-      ErasureCodingWorker.ongoingRepairs.remove(getBlockGroup().getBlockId());
+      try {
+        Thread.sleep(5000);
+      } catch (Exception e) {
+        LOG.error("Error while sleeping", e);
+      }
+
+      // ErasureCodingWorker.ongoingRepairs.remove(getBlockGroup().getBlockId());
     }
   }