[BUG] Error gives inaccurate and incomplete list of valid values when bad index.codec setting is provided

[peterskim12]

Describe the bug

When I try creating an index and pass in an invalid index.codec setting, the error returns:

{
  "error": {
    "root_cause": [
      {
        "type": "illegal_argument_exception",
        "reason": "unknown value for [index.codec] must be one of [default, lz4, best_compression, zlib] but was: zstd_no_dic"
      }
    ],
    "type": "illegal_argument_exception",
    "reason": "unknown value for [index.codec] must be one of [default, lz4, best_compression, zlib] but was: zstd_no_dic"
  },
  "status": 400
}

According to the documentation, this is not the list of valid values accepted by the API in the latest version. https://opensearch.org/docs/2.17/install-and-configure/configuring-opensearch/index-settings/#static-index-level-index-settings

It should be:

• default
• best_compression
• zstd
• zstd_no_dict
• qat_lz4 (on supported systems)
• qat_deflate (on supported systems)

I tried creating an index using index.codec = lz4 and zlib and they worked, even though the documentation does not state they are valid values. Regardless, I think the error message should be updated to match the documentation.

Related component

Indexing

To Reproduce

PUT test
{
  "settings": {
    "index.codec": "invalid_codec"
  }
}

Expected behavior

Return error with valid index.codec options:

{
  "error": {
    "root_cause": [
      {
        "type": "illegal_argument_exception",
        "reason": "unknown value for [index.codec] must be one of [default, best_compression, zstd, zstd_no_dict, qat_lz4, qat_deflate] but was: invalid_codec"
      }
    ],
    "type": "illegal_argument_exception",
    "reason": "unknown value for [index.codec] must be one of [default, best_compression, zstd, zstd_no_dict, qat_lz4, qat_deflate] but was: invalid_codec"
  },
  "status": 400
}

Additional Details

Host/Environment (please complete the following information):

• OS: macos Docker
• Version 2.19

same behavior on Fedora 39, version 2.17.1

Additional context

I can submit a PR, but I'm just following the contributing guide which says I should file an issue first. Thanks!

[dbwiddis]

Interesting. The error message is hardcoded but clearly is incomplete as it should include "lucene_default" as well as everything in Codec.availableCodecs().

OpenSearch/server/src/main/java/org/opensearch/index/engine/EngineConfig.java

Lines 135 to 162 in 9bef705

	public static final Setting<String> INDEX_CODEC_SETTING = new Setting<>("index.codec", "default", s -> {
	switch (s) {
	case "default":
	case "lz4":
	case "best_compression":
	case "zlib":
	case "lucene_default":
	return s;
	default:
	if (Codec.availableCodecs().contains(s)) {
	return s;
	}
	for (String codecName : Codec.availableCodecs()) {
	Codec codec = Codec.forName(codecName);
	if (codec instanceof CodecAliases) {
	CodecAliases codecWithAlias = (CodecAliases) codec;
	if (codecWithAlias.aliases().contains(s)) {
	return s;
	}
	}
	}
	throw new IllegalArgumentException(
	"unknown value for [index.codec] must be one of [default, lz4, best_compression, zlib] but was: " + s
	);
	}
	}, Property.IndexScope, Property.NodeScope);

I can submit a PR, but I'm just following the contributing guide which says I should file an issue first. Thanks!

Looking forward to the PR!