Skip to content

Add configurable truncation for string columns#19146

Open
jaykanakiya wants to merge 6 commits intoapache:masterfrom
jaykanakiya:string-truncation
Open

Add configurable truncation for string columns#19146
jaykanakiya wants to merge 6 commits intoapache:masterfrom
jaykanakiya:string-truncation

Conversation

@jaykanakiya
Copy link
Contributor

@jaykanakiya jaykanakiya commented Mar 12, 2026

Summary

Adds a configurable maximum string length for string dimension columns. Strings exceeding the limit are truncated during ingestion.

  • Global config: druid.indexing.formats.maxStringLength
  • Per-dimension override: maxStringLength field in the dimension spec

Release note

Added a new maxStringLength configuration for string dimensions that truncates values exceeding the specified length during ingestion. Can be set globally via druid.indexing.formats.maxStringLength or per-dimension in the ingestion spec.


Key changed/added classes in this PR
  • DefaultColumnFormatConfig
  • StringDimensionSchema
  • StringDimensionHandler
  • StringDimensionIndexer

This PR has:

  • been self-reviewed.
  • added documentation for new or modified features or behaviors.
  • a release note entry in the PR description.
  • added or updated version, license, or notice information in licenses.yaml
  • added unit tests or modified existing tests to cover new code paths, ensuring the threshold for code coverage is met.
  • added integration tests.
  • been tested in a test Druid cluster.

Comment on lines +54 to +55
@JsonProperty("createBitmapIndex") Boolean createBitmapIndex,
@JsonProperty("maxStringLength") @Nullable Integer maxStringLength
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

drive by comment (i'll have a closer look at rest of PR later)

instead of adding additional arguments here, I was hoping to deprecate these arguments in favor of adding a column format spec similar to was done for auto/json columns in #17762, which could serve as a reference for how this should be wired up. I was planning to move the existing createBitmapIndex and multiValueHandling into such a spec, but just haven't got to it yet. I think this would be much cleaner and less disruptive to call sites going forward. It also allows wiring up to IndexSpec to be able to define job level defaults as a middle place between per column and system wide.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for taking a look @clintropolis. Adding something like StringCommonFormatColumnFormatSpec would make it cleaner and makes sense to consolidate the configs there. Since it seems like a bigger refactor, does it make sense to do it in a follow up? Let me know what you think.

*/
private static DimensionSchema.MultiValueHandling STRING_MV_MODE = DimensionSchema.MultiValueHandling.SORTED_ARRAY;
private static IndexSpec DEFAULT_INDEX_SPEC = IndexSpec.builder().build();
private static int MAX_STRING_LENGTH = 0;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would it make sense to set this to Integer max value? In case this is used elsewhere in the future there wouldn't need to explicit handling for 0 like you have in truncateIfNeeded

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We are using NON_DEFAULT to not serialize the default value and I think for integer jackson's default is 0. If we set the MAX_STRING_LENGTH default as int max, it'll serialize this value for each dimension.

public SideEffectRegisterer initDimensionHandlerAndMvHandlingMode(DefaultColumnFormatConfig formatsConfig)
{
setStringMultiValueHandlingModeIfConfigured(formatsConfig.getStringMultiValueHandlingMode());
setMaxStringLengthIfConfigured(formatsConfig.getMaxStringLength());
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can take a look at druid.indexing.formats.stringMultiValueHandlingMode in BuiltInTypesModuleTest It would be good to have some test coverage for the new property

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Added tests for this property.

this.stringMultiValueHandlingMode = validateMultiValueHandlingMode(stringMultiValueHandlingMode);
this.nestedColumnFormatVersion = nestedColumnFormatVersion;
this.indexSpec = indexSpec;
this.maxStringLength = maxStringLength;
Copy link
Contributor

@aho135 aho135 Mar 19, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Here we should validate that configured maxStringLength > 0, otherwise we can throw an exception or log that we are falling back to the default value

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Added validations.

);
}

private long verifyEncodedValues(
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Leaving a comment for us to revisit later:

I'm curious what the truncation behavior would be like for MVD

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I did have a chat with @abhishekrb19 about multi value strings and we decided to leave out truncation for those.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ahh okay. I see we're truncating for MVD's with a single value. Would be good to add a test case for that also if that's the intended behavior


@AfterEach
public void beforeEach()
@After
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think there's some mix of junit 4/5 annotations due to which AfterEach was not firing. Updated to use junit 4's annotation for clean up.

Copy link
Contributor

@aho135 aho135 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the contribution @jaykanakiya! The changes look good to me. I'll let @clintropolis take a pass through as well. I think it would be fine to do the suggested refactoring in a follow up

);
}

private long verifyEncodedValues(
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ahh okay. I see we're truncating for MVD's with a single value. Would be good to add a test case for that also if that's the intended behavior

|`druid.indexer.task.tmpStorageBytesPerTask`|Maximum number of bytes per task to be used to store temporary files on disk. This config is generally intended for internal usage. Attempts to set it are very likely to be overwritten by the TaskRunner that executes the task, so be sure of what you expect to happen before directly adjusting this configuration parameter. The config is documented here primarily to provide an understanding of what it means if/when someone sees that it has been set. A value of -1 disables this limit. |-1|
|`druid.indexer.task.allowHadoopTaskExecution`|Conditional dictating if the cluster allows `index_hadoop` tasks to be executed. `index_hadoop` is deprecated, and defaulting to false will force cluster operators to acknowledge the deprecation and consciously opt in to using index_hadoop with the understanding that it will be removed in the future.|false|
|`druid.indexer.server.maxChatRequests`|Maximum number of concurrent requests served by a task's chat handler. Set to 0 to disable limiting.|0|
|`druid.indexing.formats.maxStringLength`|Maximum number of characters to store per string dimension value. Longer values are truncated during ingestion. Set to 0 to disable. Can be overridden per-dimension using `maxStringLength` in the [dimension object](../ingestion/ingestion-spec.md#dimension-objects).|0 (no truncation)|
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It would be good to mention that truncation does not apply for MVD's

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I added the test and doc update in the next commit.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants