prestodb/presto · error · PrestoException

HIVE_BAD_DATA

HIVE_BAD_DATA

Error message

Error parsing symlinks

What it means

QuickStatsProvider.buildQuickStats reads symlink text files to enumerate a partition's data files; when the underlying FileSystem/LocalFileSystem throws IOException while parsing those symlink targets, the provider wraps it in a PrestoException with error code HIVE_BAD_DATA. It signals the partition's symlink data is malformed or unreadable, not a transient failure.

Source

Thrown at presto-hive/src/main/java/com/facebook/presto/hive/statistics/QuickStatsProvider.java:389

                for (Map.Entry<Path, List<Path>> entry : parentToTargets.entrySet()) {
                    targetFileInfoList.addAll(getTargetPathsHiveFileInfos(
                            path,
                            partition,
                            entry.getKey(),
                            entry.getValue(),
                            hiveDirectoryContext,
                            fs,
                            directoryLister,
                            resolvedTable,
                            nameNodeStats,
                            session));
                }

                fileList = targetFileInfoList.build().iterator();
            }
            catch (IOException e) {
                throw new PrestoException(HIVE_BAD_DATA, "Error parsing symlinks", e);
            }
        }

        PartitionQuickStats partitionQuickStats = PartitionQuickStats.EMPTY;
        Stopwatch buildStopwatch = Stopwatch.createStarted();
        // Build quick stats one by one from statsBuilderStrategies. Do this until we get a non-empty PartitionQuickStats
        for (QuickStatsBuilder strategy : statsBuilderStrategies) {
            partitionQuickStats = strategy.buildQuickStats(session, metastore, table, metastoreContext, partitionId, fileList);

            if (partitionQuickStats != PartitionQuickStats.EMPTY) {
                // Strategy successfully resolved stats, don't explore other strategies
                // TODO : We can order the strategies based on table metadata, e.g Iceberg tables could use the IcebergQuickStatsBuilder first
                break;
            }
        }

        long buildMillis = buildStopwatch.elapsed(MILLISECONDS);
        session.getRuntimeStats().addMetricValue("QuickStatsProvider/BuildTimeMS/" + partitionKey, RuntimeUnit.NONE, buildMillis);

View on GitHub (pinned to 55bb57d202)

Solutions

  1. Inspect and fix the symlink text file for the partition (remove/repair dangling or malformed entries)
  2. Verify all symlink target paths exist and are readable by the Presto user
  3. Check the underlying filesystem health/permissions (HDFS NameNode logs, hdfs fsck)
  4. Recreate the partition with a correct LOCATION/symlink metadata, e.g. via MSCK or ALTER TABLE ... ADD PARTITION
  5. Look at the wrapped IOException (cause) for the precise filesystem error

Example fix

// before (bad symlink file entry)
hdfs://namenode/missing/path/file.orc
// after (fixed symlink entry)
hdfs://namenode/data/table/part=2026-01-01/file.orc
Defensive patterns

Strategy: try-catch

Validate before calling

// before buildQuickStats: validate the symlink file
for (String line : symlinkFileLines) {
    Path target = new Path(line.trim());
    if (!fileSystem.exists(target)) {
        throw new PrestoException(HIVE_BAD_DATA, "Symlink target missing: " + target);
    }
}

Try / catch

try {
    stats = quickStatsProvider.buildQuickStats(...);
} catch (PrestoException e) {
    if (HIVE_BAD_DATA.toErrorCode().equals(e.getErrorCode())) {
        LOG.error("Symlink partition data invalid: " + e.getMessage(), e);
        // surface to user / fail query with a clear message
    }
    throw e;
}

Prevention

When it happens

Trigger: A Hive partition defined with symlink-style location whose symlink file contains bad paths, unreadable targets, or the underlying filesystem raises IOException during targetFileInfo resolution.

Common situations: Manually edited symlink text files with dangling or malformed targets; permissions/missing files referenced by symlinks; partitions moved or renamed without updating symlink files; HDFS outages during listing.

Related errors

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/7224f42a6b063b02. Report an issue: GitHub.