[jira] [Created] (HUDI-5496) Prevent Hudi from generating clustering plans with filegroups consisting of only 1 fileSlice

["voon (Jira)" <ji...@apache.org>]

voon created HUDI-5496:
--------------------------

             Summary: Prevent Hudi from generating clustering plans with filegroups consisting of only 1 fileSlice
                 Key: HUDI-5496
                 URL: https://issues.apache.org/jira/browse/HUDI-5496
             Project: Apache Hudi
          Issue Type: Bug
            Reporter: voon


Suppose a partition is no longer being written/updated, i.e. there will be no changes to the partition, therefore, size of parquet files will always be the same. 

 

If the parquet files in the partition (even after prior clustering) is smaller than {*}hoodie.clustering.plan.strategy.small.file.limit{*}, ** the fileSlice will always be returned as a candidate for {_}getFileSlicesEligibleForClustering(){_}.

 

This may cause inputGroups with only 1 fileSlice to be selected as candidates for clustering. An of a clusteringPlan demonstrating such a case in JSON format is seen below.

 

 
{code:java}
{
  "inputGroups": [
    {
      "slices": [
        {
          "dataFilePath": "/path/clustering_test_table/dt=2023-01-03/cf2929a7-78dc-4e99-be0c-926e9487187d-0_0-2-0_20230104102201656.parquet",
          "deltaFilePaths": [],
          "fileId": "cf2929a7-78dc-4e99-be0c-926e9487187d-0",
          "partitionPath": "dt=2023-01-03",
          "bootstrapFilePath": "",
          "version": 1
        }
      ],
      "metrics": {
        "TOTAL_LOG_FILES": 0.0,
        "TOTAL_IO_MB": 260.0,
        "TOTAL_IO_READ_MB": 130.0,
        "TOTAL_LOG_FILES_SIZE": 0.0,
        "TOTAL_IO_WRITE_MB": 130.0
      },
      "numOutputFileGroups": 1,
      "extraMetadata": null,
      "version": 1
    },
    {
      "slices": [
        {
          "dataFilePath": "/path/clustering_test_table/dt=2023-01-04/b101162e-4813-4de6-9881-4ee0ff918f32-0_0-2-0_20230104103401458.parquet",
          "deltaFilePaths": [],
          "fileId": "b101162e-4813-4de6-9881-4ee0ff918f32-0",
          "partitionPath": "dt=2023-01-04",
          "bootstrapFilePath": "",
          "version": 1
        },
        {
          "dataFilePath": "/path/clustering_test_table/dt=2023-01-04/9b1c1494-2a58-43f1-890d-4b52070937b1-0_0-2-0_20230104102201656.parquet",
          "deltaFilePaths": [],
          "fileId": "9b1c1494-2a58-43f1-890d-4b52070937b1-0",
          "partitionPath": "dt=2023-01-04",
          "bootstrapFilePath": "",
          "version": 1
        }
      ],
      "metrics": {
        "TOTAL_LOG_FILES": 0.0,
        "TOTAL_IO_MB": 418.0,
        "TOTAL_IO_READ_MB": 209.0,
        "TOTAL_LOG_FILES_SIZE": 0.0,
        "TOTAL_IO_WRITE_MB": 209.0
      },
      "numOutputFileGroups": 1,
      "extraMetadata": null,
      "version": 1
    }
  ],
  "strategy": {
    "strategyClassName": "org.apache.hudi.client.clustering.run.strategy.SparkSortAndSizeExecutionStrategy",
    "strategyParams": {},
    "version": 1
  },
  "extraMetadata": {},
  "version": 1,
  "preserveHoodieMetadata": true
}{code}
 

Such a case will cause performance issues as a parquet file is re-written unnecessarily (write amplification). 

 

The fix is to only select inputGroups with more than 1 fileSlice as candidates for clustering.

 



--
This message was sent by Atlassian Jira
(v8.20.10#820010)