apache/hadoop · error · IOException

Could not find image with txid " + txid

Error message

Could not find image with txid " + txid

What it means

ImageServlet's GETIMAGE handling throws IOException("Could not find image with txid N") when the request asks for a specific transaction-id fsimage and NNStorage.getFsImage(txid, {IMAGE, IMAGE_ROLLBACK}) finds no matching fsimage_<txid> file in the NameNode's storage directories. The NameNode simply does not retain (or never had) an image at that txid.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/ImageServlet.java:158

      validateRequest(context, conf, request, response, nnImage,
          parsedParams.getStorageInfoString());

      UserGroupInformation.getCurrentUser().doAs(new PrivilegedExceptionAction<Void>() {
        @Override
        public Void run() throws Exception {
          if (parsedParams.isGetImage()) {
            long txid = parsedParams.getTxId();
            File imageFile = null;
            String errorMessage = "Could not find image";
            if (parsedParams.shouldFetchLatest()) {
              imageFile = nnImage.getStorage().getHighestFsImageName();
            } else {
              errorMessage += " with txid " + txid;
              imageFile = nnImage.getStorage().getFsImage(txid,
                  EnumSet.of(NameNodeFile.IMAGE, NameNodeFile.IMAGE_ROLLBACK));
            }
            if (imageFile == null) {
              throw new IOException(errorMessage);
            }
            CheckpointFaultInjector.getInstance().beforeGetImageSetsHeaders();
            long start = monotonicNow();
            serveFile(imageFile);

            if (metrics != null) { // Metrics non-null only when used inside name node
              long elapsed = monotonicNow() - start;
              metrics.addGetImage(elapsed);
            }
          } else if (parsedParams.isGetEdit()) {
            long startTxId = parsedParams.getStartTxId();
            long endTxId = parsedParams.getEndTxId();
            
            File editFile = nnImage.getStorage()
                .findFinalizedEditsFile(startTxId, endTxId);
            long start = monotonicNow();
            serveFile(editFile);

View on GitHub (pinned to 2add963021)

Solutions

1. Request the latest image instead of a fixed txid (GetImageParams 'latest' flag, e.g. getimage?getimage=1&latest=1)
2. Verify which images exist: ls <dfs.namenode.name.dir>/current/fsimage_* or check NameNode JMX mostRecentCheckpointTxid
3. Re-run the checkpoint or bootstrap so the requester and the NameNode agree on the current txid
4. Increase dfs.namenode.num.checkpoints.retained if requesters legitimately need older checkpoints

Example fix

# before: pinned txid that was purged
curl -f 'http://nn:9870/getimage?getimage=1&txid=1234'
# -> IOException: Could not find image with txid 1234

# after: fetch latest
TXID=$(curl -s http://nn:9870/jmx | jq -r '.beans[]|select(.name=="Hadoop:service=NameNode,name=NameNodeInfo").MostRecentCheckpointTxId')
curl -f "http://nn:9870/getimage?getimage=1&txid=$TXID"

// Before requesting a specific txid, confirm the NN retains it
long latest = jmxMostRecentCheckpointTxid(nnJmxUrl);   // NameNodeInfo bean
if (requestedTxid > latest) {
  requestedTxid = latest;   // or fail: request is for a purged checkpoint
}
curl -f "http://nn:9870/getimage?getimage=1&txid=$requestedTxid"

catch (IOException e) {
  if (e.getMessage() != null && e.getMessage().contains("Could not find image with txid")) {
    imageFile = fetchLatest();   // fall back to latest=1 transfer
    if (imageFile == null) { throw e; }  // NN storage genuinely inconsistent
  } else { throw e; }
}

When it happens

Trigger: Requesting a checkpoint txid already purged by retention (dfs.namenode.num.checkpoints.retained); asking for an IMAGE_ROLLBACK file when no rollback checkpoint exists; SecondaryNameNode/Standby bootstrapping with a stale txid after the Active's storage changed (restart, fresh format, restored directory).

Common situations: SNN checkpoint failing right after NN storage was replaced or reformatted; too few retained checkpoints for a slow/down SNN; hand-crafted transfer URLs or scripts pinning an old txid.

Related errors

• The length of the feature flag section was negative at {} by
• Content-Length header is not provided by the namenode when t
• Safe mode should be turned ON in order to create namespace i
• NameNode initialization not yet complete. FSImage has not be
• GetImage failed. " + StringUtils.stringifyException(t)

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/6f3bf36ce7a62e4c. Report an issue: GitHub.