apache/hadoop · error · IOException

block pool {bpid} is not found

Error message

block pool {bpid} is not found

What it means

FsVolumeImpl.getBlockPoolSlice(bpid) returns this volume's per-block-pool storage; the IOException fires when the bpid is absent from bpSlices - the volume has not (yet) created the block pool directory, or the pool was already removed. Operations touching blocks of a freshly registered or freshly retired NameNode hit this race.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/fsdataset/impl/FsVolumeImpl.java:565

  @VisibleForTesting
  long getRecentReserved() {
    return recentReserved;
  }

  public Map<String, BlockPoolSlice> getBlockPoolSlices() {
    return bpSlices;
  }

  long getReserved(){
    return reserved != null ? reserved.getReserved() : 0;
  }

  @VisibleForTesting
  BlockPoolSlice getBlockPoolSlice(String bpid) throws IOException {
    BlockPoolSlice bp = bpSlices.get(bpid);
    if (bp == null) {
      throw new IOException("block pool " + bpid + " is not found");
    }
    return bp;
  }

  @Override
  public URI getBaseURI() {
    return baseURI;
  }

  @Override
  public DF getUsageStats(Configuration conf) {
    if (currentDir != null) {
      try {
        return new DF(new File(currentDir.getParent()), conf);
      } catch (IOException e) {
        LOG.error("Unable to get disk statistics for volume {}", this, e);
      }
    }

View on GitHub (pinned to 2add963021)

Solutions

1. Retry after registration settles - addBlockPool creates the slice during handshake; watch for 'Adding block pool' in the DN log.
2. Verify the NN's cluster ID and block pool ID match the DN's; a mismatched VERSION prevents BP creation.
3. If the BP was retired, ensure clients stop referencing it (refresh nameservices on clients).
4. Persistent occurrence with a valid bpid means the BP dir failed to create - check permissions and space under the volume.

if (!volume.getBlockPoolSlices().containsKey(bpid)) {
  // BP slice not ready or removed: retry after registration completes
  return;
}
BlockPoolSlice slice = volume.getBlockPoolSlice(bpid);

try {
  BlockPoolSlice bp = volume.getBlockPoolSlice(bpid);
} catch (IOException e) {
  if (e.getMessage().contains("is not found")) {
    // wait for 'Adding block pool' then retry
  }
}

When it happens

Trigger: A block request arrives between NN handshake and addBlockPool creating the BP slice; refreshNamenodes adds a nameservice whose slice is still initializing; the BP was deleted but a stale request still references it.

Common situations: DN restart while jobs run; adding or removing nameservices; transient errors right after NameNode registration; usually retried by the caller.

Related errors

• Not ready to serve the block pool, {}.
• DN shut down before block pool registered
• Storage directory for location {} and block pool id {} does
• No block pool offer service for bpid={}
• FsDatasetSpi has not been initialized

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/b75ee06c40a689b0. Report an issue: GitHub.