From 93191b4da0fea0d41b7d5e58269ea58c6e24ef58 Mon Sep 17 00:00:00 2001 From: wangd Date: Thu, 6 Feb 2025 08:51:11 +0800 Subject: [PATCH] [ToBeSquashed]Fix document --- .../src/main/sphinx/connector/iceberg.rst | 23 +++++++++++-------- 1 file changed, 14 insertions(+), 9 deletions(-) diff --git a/presto-docs/src/main/sphinx/connector/iceberg.rst b/presto-docs/src/main/sphinx/connector/iceberg.rst index a45828396e881..8287a60e82ae3 100644 --- a/presto-docs/src/main/sphinx/connector/iceberg.rst +++ b/presto-docs/src/main/sphinx/connector/iceberg.rst @@ -256,14 +256,26 @@ Property Name Description ======================================================= ============================================================= ============ ``iceberg.catalog.warehouse`` The catalog warehouse root path for Iceberg tables. + Hadoop catalog requires a file system that supports atomic + rename operation, such as HDFS, to maintain metadata files + in order to implement atomic transaction commit. + Example: ``hdfs://nn:8020/warehouse/path`` + + Do not set ``iceberg.catalog.warehouse`` to a path in object + stores or local file systems in the production environment. + This property is required if the ``iceberg.catalog.type`` is ``hadoop``. ``iceberg.catalog.hadoop.warehouse.datadir`` The catalog warehouse root data path for Iceberg tables. + Its only supported with Hadoop catalog. + Example: ``s3://iceberg_bucket/warehouse``. - This property is only supported with Hadoop catalog. If set, - all tables in this Hadoop catalog will default to saving + + This optional property can be set to a path in object + stores or HDFS. + If set, all tables in this Hadoop catalog default to saving their data and delete files in the specified root data directory. @@ -272,13 +284,6 @@ Property Name Description Otherwise, it will be ignored. ======================================================= ============================================================= ============ -.. note:: - - Hadoop catalog requires a file system that supports atomic rename operation, such as HDFS, - to maintain metadata files in order to implement atomic transaction commit. It's unsafe to - set `iceberg.catalog.warehouse` to a path in object stores and local file systems. But it's - safe to set `iceberg.catalog.hadoop.warehouse.datadir` to a path in the object stores. - Configure the `Amazon S3 `_ properties to specify a S3 location as the warehouse data directory for the Hadoop catalog. This way, the data and delete files of Iceberg tables are stored in S3. An example configuration includes: