README.md in embulk-output-redshift-0.6.0 vs README.md in embulk-output-redshift-0.6.1

@@ -21,11 +21,12 @@
 - **secret_access_key**: secret access key for AWS
 - **iam_user_name**: IAM user name for uploading temporary files to S3. The user should have permissions of `s3:GetObject`, `s3:PutObject`, `s3:DeleteObject`, , `s3:ListBucket` and `sts:GetFederationToken`. (string, default: "", but we strongly recommend that you use IAM user for security reasons. see below.)
 - **s3_bucket**: S3 bucket name for temporary files
 - **s3_key_prefix**: S3 key prefix for temporary files (string, default:"")
 - **options**: extra connection properties (hash, default: {})
-- **mode**: "insert", "insert_direct", "truncate_insert", or "replace". See below. (string, required)
+- **mode**: "insert", "insert_direct", "truncate_insert", "replace" or "merge". See below. (string, required)
+- **merge_keys**: key column names for merging records in merge mode (string array, required in merge mode)
 - **batch_size**: size of a single batch insert (integer, default: 16777216)
 - **default_timezone**: If input column type (embulk type) is timestamp, this plugin needs to format the timestamp into a SQL string. This default_timezone option is used to control the timezone. You can overwrite timezone for each columns using column_options option. (string, default: `UTC`)
 - **column_options**: advanced: a key-value pairs where key is a column name and value is options for the column.
   - **type**: type of a column when this plugin creates new tables (e.g. `VARCHAR(255)`, `INTEGER NOT NULL UNIQUE`). This used when this plugin creates intermediate tables (insert, truncate_insert and merge modes), when it creates the target table (insert_direct and replace modes), and when it creates nonexistent target table automatically. (string, default: depends on input column type. `BIGINT` if input column type is long, `BOOLEAN` if boolean, `DOUBLE PRECISION` if double, `CLOB` if string, `TIMESTAMP` if timestamp)
   - **value_type**: This plugin converts input column type (embulk type) into a database type to build a INSERT statement. This value_type option controls the type of the value in a INSERT statement. (string, default: depends on the sql type of the column. Available values options are: `byte`, `short`, `int`, `long`, `double`, `float`, `boolean`, `string`, `nstring`, `date`, `time`, `timestamp`, `decimal`, `json`, `null`, `pass`)
@@ -49,9 +50,13 @@
   * Resumable: Yes.
 * **replace**:
   * Behavior: This mode writes rows to an intermediate table first. If all those tasks run correctly, drops the target table and alters the name of the intermediate table into the target table name.
   * Transactional: Yes.
   * Resumable: No.
+* **merge**:
+  * Behavior: This mode writes rows to some intermediate tables first. If all those tasks run correctly, inserts records from intermediate tables after deleting records whose keys exist in intermediate tables. Namely, if merge keys of a record in the intermediate tables already exist in the target table, the target record is updated by the intermediate record, otherwise the intermediate record is inserted. If the target table doesn't exist, it is created automatically.
+  * Transactional: Yes.
+  * Resumable: Yes.
 
 ### Supported types
 
 |database type|default value_type|note|
 |:--|:--|:--|
\ No newline at end of file