prestodb/presto · error · PrestoException

HIVE_PATH_ALREADY_EXISTS

HIVE_PATH_ALREADY_EXISTS

Error message

Unable to create directory %s: target directory already exists

What it means

During commit of a semi-transactional session, a staging/target directory rename or create finds the target path already exists when it must be fresh; the HDFS directory creation helper refuses to overwrite and reports the target path, preventing duplicate or clobbered data directories.

Source

Thrown at presto-hive-metastore/src/main/java/com/facebook/presto/hive/metastore/SemiTransactionalHiveMetastore.java:1412

                            context,
                            hdfsEnvironment,
                            currentPath.get(),
                            targetPath,
                            () -> cleanUpTasksForAbort.add(new DirectoryCleanUpTask(context, targetPath, true)));
                }
                else {
                    // CREATE TABLE AS SELECT partitioned table, or
                    // CREATE TABLE AS SELECT unpartitioned table without temporary staging directory
                    // CREATE TABLE partitioned/unpartitioned table (without data)
                    if (pathExists(context, hdfsEnvironment, targetPath)) {
                        if (currentPath.isPresent() && currentPath.get().equals(targetPath)) {
                            // It is okay to skip directory creation when currentPath is equal to targetPath
                            // because the directory may have been created when creating partition directories.
                            // However, it is important to note that the two being equal does not guarantee
                            // a directory had been created.
                        }
                        else {
                            throw new PrestoException(
                                    HIVE_PATH_ALREADY_EXISTS,
                                    format("Unable to create directory %s: target directory already exists", targetPath));
                        }
                    }
                    else {
                        cleanUpTasksForAbort.add(new DirectoryCleanUpTask(context, targetPath, true));
                        createDirectory(context, hdfsEnvironment, targetPath);
                    }
                }
            }
            addTableOperations.add(new CreateTableOperation(metastoreContext, table, tableAndMore.getPrincipalPrivileges(), tableAndMore.isIgnoreExisting(), tableAndMore.getConstraints()));
            if (!isPrestoView(table)) {
                updateStatisticsOperations.add(new UpdateStatisticsOperation(metastoreContext,
                        table.getSchemaTableName(),
                        Optional.empty(),
                        tableAndMore.getStatisticsUpdate(),
                        false));
            }

View on GitHub (pinned to 55bb57d202)

Solutions

1. Remove stale staging/target directories left over from a previous failed query, then retry
2. Ensure concurrent queries do not commit the same table/partition target paths simultaneously
3. Check for orphaned directories from aborted transactions and clean them via cleanup tooling

When it happens

Trigger: Thrown at presto-hive-metastore/src/main/java/com/facebook/presto/hive/metastore/SemiTransactionalHiveMetastore.java:1412 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of prestodb/presto@55bb57d202 (2026-09-04). Data as JSON: /api/errors/3580a63651c96a6e. Report an issue: GitHub.