magda-pdf-semantic-indexer

A Helm chart for Magda PDF Semantic Indexer

Homepage: https://github.com/magda-io/magda-pdf-semantic-indexer

Source Code

• https://github.com/magda-io/magda-pdf-semantic-indexer

Requirements

Kubernetes: >= 1.14.0-0

Repository	Name	Version
oci://ghcr.io/magda-io/charts	magda-common	5.2.0

Values

Key	Type	Default	Description
defaultAdminUserId	string	"00000000-0000-4000-8000-000000000000"
defaultImage.imagePullSecret	bool	false
defaultImage.pullPolicy	string	"IfNotPresent"
defaultImage.repository	string	"ghcr.io/magda-io"
defaultSemanticIndexerConfig.bulkEmbeddingsSize	int	1
defaultSemanticIndexerConfig.bulkIndexSize	int	50
defaultSemanticIndexerConfig.chunkSizeLimit	int	512
defaultSemanticIndexerConfig.id	string	"pdf-semantic-indexer"
defaultSemanticIndexerConfig.indexName	string	"semantic-index"
defaultSemanticIndexerConfig.indexVersion	int	1
defaultSemanticIndexerConfig.overlap	int	50
embeddingApiURL	string	"http://magda-embedding-api"
global	object	{"image":{},"rollingUpdate":{},"searchEngine":{"defaultDatasetBucket":"magda-datasets","semanticIndexer":{"indexName":null,"indexVersion":null,"knnVectorFieldConfig":{"compressionLevel":"32x","dimension":768,"efConstruction":100,"efSearch":100,"m":16,"mode":"on_disk","spaceType":"l2"},"numberOfReplicas":0,"numberOfShards":1}}}	only for providing appropriate default value for helm lint
global.searchEngine.semanticIndexer.knnVectorFieldConfig.compressionLevel	string	"32x"	The compression_level mapping parameter selects a quantization encoder that reduces vector memory consumption by the given factor.
global.searchEngine.semanticIndexer.knnVectorFieldConfig.dimension	int	768	Dimension of the embedding vectors.
global.searchEngine.semanticIndexer.knnVectorFieldConfig.efConstruction	int	100	Similar to efSearch but used during index construction. Higher values improve search quality but increase index build time.
global.searchEngine.semanticIndexer.knnVectorFieldConfig.efSearch	int	100	The size of the candidate queue during search. Larger values may improve search quality but increase search latency.
global.searchEngine.semanticIndexer.knnVectorFieldConfig.m	int	16	The maximum number of graph edges per vector. Higher values increase memory usage but may improve search quality.
global.searchEngine.semanticIndexer.knnVectorFieldConfig.mode	string	"on_disk"	Vector workload mode: on_disk or in_memory.
image.name	string	"magda-pdf-semantic-indexer"
livenessProbe	object	{"initialDelaySeconds":10,"periodSeconds":10,"timeoutSeconds":60}	Liveness probe settings.
livenessProbe.initialDelaySeconds	int	10	Seconds after container start before probes begin.
livenessProbe.periodSeconds	int	10	Seconds between liveness probes.
livenessProbe.timeoutSeconds	int	60	Probe timeout in seconds.
minioConfig.defaultDatasetBucket	string	""
minioConfig.endPoint	string	"magda-minio"
minioConfig.port	int	9000
minioConfig.region	string	""
minioConfig.useSSL	bool	false
opensearchURL	string	"http://opensearch:9200"
port	int	6305	Service port configuration
resources	object	{"limits":{"cpu":"1000m","memory":"1024Mi"},"requests":{"cpu":"100m","memory":"200Mi"}}	Container resources for the indexer pod.
semanticIndexer.bulkEmbeddingsSize	int	nil	number of string we request embedding api to process in one request
semanticIndexer.bulkIndexSize	int	nil	Number of documents we send to OpenSearch for bulk processing in a single request
semanticIndexer.chunkSizeLimit	int	nil	The maximum number of tokens in a single chunk.
semanticIndexer.id	string	nil	Semantic indexer ID
semanticIndexer.indexName	string	nil	index name
semanticIndexer.indexVersion	int	nil	index version
semanticIndexer.overlap	int	nil	The number of overlapping tokens between chunks.

---

Autogenerated from chart metadata using helm-docs v1.11.0