GMS lite

Run the production war.war locally without Docker, Kafka, or a database server. Requires Python 3.10+, a POSIX system (Linux or macOS), and JDK 25. Python uses only the standard library. The first Gradle build downloads JVM dependencies.

export JAVA_HOME=~/.local/share/mise/installs/java/25.0.2
python3 metadata-service/gms-lite/bin/gms-lite-stack.py

The launcher builds the GMS and upgrade jars plus the embedded OpenSearch installDist, gets or starts the shared OpenSearch node, then runs datahub-upgrade.jar -u SystemUpdate and -u RestoreIndices for each new instance. Upgrade processes run serially to limit memory use. GMS processes start after bootstrap finishes. Readiness requires HTTP health and, when authentication is enabled, the bootstrap user's policies.

Use --skip-build after building those artifacts. --heap 1g sets each GMS/upgrade heap cap. One OpenSearch JVM, with a 4 GB default heap, serves all local stacks. Set EMBEDDED_OPENSEARCH_HEAP when building its distribution to change that cap, then stop and start the shared node to apply it:

EMBEDDED_OPENSEARCH_HEAP=6g ./gradlew :metadata-service:opensearch-embedded:installDist
python3 metadata-service/gms-lite/bin/_opensearch_manager.py stop
python3 metadata-service/gms-lite/bin/_opensearch_manager.py start

The shared node keeps its data, logs, port, pid, and startup.lock under ~/.datahub/local/opensearch-3/. This versioned directory separates major versions because OpenSearch data directories are not portable across major versions. Override it with --shared-opensearch-dir /path/to/state on the launcher or GMS_LITE_OPENSEARCH_DIR for both entry points. Manager commands accept --dir /path/to/state. Always use the same directory to manage a node.

The manager serializes start/stop with an exclusive file lock, verifies process identity before signalling, and reuses only a healthy OpenSearch 3.x node. A dead node or a verified node whose connection is refused is replaced while retaining its data; a different major version or an unrelated PID is refused. For a live node, readiness waits up to 600 seconds for yellow health. Override this with GMS_LITE_OPENSEARCH_HEALTH_TIMEOUT_SECONDS. An unresponsive or slow node that reaches the deadline causes a clear error and is left running, with its recorded PID and port unchanged. This also applies to a newly started node: once it reports READY, its port is recorded before waiting for yellow. Retry start to reuse it and wait again. A child that exits, fails to report READY within 900 seconds, or reports the wrong major version is stopped and its PID/port files are removed. Interrupting startup also cleans up the new child; an already running node being reused is left alone. Data is always retained; choose a different state directory or repair the data explicitly if it cannot recover. status snapshots PID/port under the lock, then probes outside it with one-second HTTP timeouts and no readiness wait. It returns JSON with healthy: false and a short reason for a red, paused, unreachable, or incompatible node. There is no automatic deletion of other stacks' indices.

Both HTTP and transport ports are ephemeral. Use --private-opensearch for a node owned by this stack, with data under <data-dir>/opensearch/, that stops with the launcher. Supplying --opensearch-port N also selects private mode (0 means ephemeral); occupied ports are refused. Private mode is useful for CI and self-contained runs.

python3 metadata-service/gms-lite/bin/gms-lite-stack.py \
  --skip-build --data-dir /tmp/my-gms --port-base 18090 --processes 2

--processes defaults to 1. Instances are named gms-0, gms-1, etc. Each has its own H2 file, queue spool, object-storage directory, HTTP port, and index prefix. Prefixes include a hash of the absolute data-directory path and the instance name, avoiding collisions between identically named directories. Instances share one OpenSearch node but never an index. By default the shared node runs in its own session and outlives every launcher that uses it; stopping one stack leaves other stacks running. The launcher prints the command to stop the shared node:

python3 metadata-service/gms-lite/bin/_opensearch_manager.py status
python3 metadata-service/gms-lite/bin/_opensearch_manager.py stop

Stopping the shared node interrupts search for every stack using it. Stop stacks first. The manager sends SIGTERM, waits for shutdown, and uses SIGKILL only if necessary. start can also prewarm the node without starting GMS.

An optional --env-file /path/to/overrides.env accepts literal KEY=VALUE lines, optional export prefixes, quoted values, and whole-line comments. There is no default env file. The launcher inherits the shell environment, applies this file, and then pins its local connection settings. Search URI, credentials, TLS, proxy, routing, and per-cluster prefix overrides are neutralised for both GMS and upgrades so the embedded node and instance prefix always win. The full variable-to-YAML mapping appears in the environment contract below. The generated gms-N/env.sh records settings that differ from the inherited environment, using shell-safe quoting. It is an output file and is never read back automatically.

Discovery and authentication

Once ready, <data-dir>/stack.json is published atomically, with mode 0600:

{
  "bootId": "unique-random-id-for-this-start",
  "opensearchPort": 54321,
  "opensearchShared": true,
  "default": "gms-0",
  "instances": {
    "gms-0": "http://localhost:18090",
    "gms-1": "http://localhost:18091"
  },
  "indexPrefixes": {
    "gms-0": "gmslite0123456789abgms0",
    "gms-1": "gmslite0123456789abgms1"
  },
  "auth": true,
  "adminTokens": {
    "gms-0": "<personal-access-token>",
    "gms-1": "<personal-access-token>"
  },
  "adminTokenExpiresAt": {
    "gms-0": "2026-01-02T12:00:00Z",
    "gms-1": "2026-01-02T12:00:01Z"
  }
}

opensearchShared says whether stopping this launcher leaves the node running. default is a key in instances. indexPrefixes contains the configured INDEX_PREFIX values; GMS appends an underscore before index names. bootId changes on every start. adminTokens contains a one-day personal token for the existing bootstrap user datahub in each instance. adminTokenExpiresAt maps the same instance names to the server-issued token expiry in ISO-8601 UTC. After expiry, authenticated requests return 401 even if /health still succeeds. Restart the stack with the same data directory to mint new tokens, then reread stack.json; tokens are not refreshed while the launcher runs. The launcher uses OSS's shipped bootstrap policies without adding or replacing any policy. No frontend or password login is started.

Pass the instance token as Authorization: Bearer <token>, or use it with the DataHub CLI as DATAHUB_GMS_TOKEN alongside DATAHUB_GMS_URL. /health and /config are public. For example:

curl -fsS http://localhost:18090/health
curl -fsS http://localhost:18090/config

--no-auth disables token authentication and publishes auth: false with empty adminTokens and adminTokenExpiresAt objects. GMS binds to loopback. The system credentials and token keys are randomly generated, saved in <data-dir>/.secrets.json with mode 0600, and reused on restart. Neither credentials nor tokens are printed to stdout.

Persistence and recovery

Stop with Ctrl-C or SIGTERM to the launcher PID. It stops GMS and active upgrade processes, waits for JVM shutdown, and kills remaining children in their owned process groups if they do not exit. Shared OpenSearch stays up. In private mode, GMS stops before the owned OpenSearch node. The spool is saved during orderly JVM shutdown; SIGKILL or power loss can lose queued events. This is a local development stack.

Restart with the same absolute data-directory path. H2, search indices, credentials, and queue snapshots persist. A gms-N/system-update-done marker is written only after both upgrade commands succeed; it records the instance's non-usage-event index names observed at that point. Existing H2 tables are never recreated on a retry. Upgrades are skipped only when H2 exists, this marker contains a nonempty recorded index set, and every recorded index still exists. If any recorded index is missing, both upgrades run again to restore search from H2. The usage-event index is ensured separately on every start. Older text markers trigger one upgrade run to record the set. Remove the marker to rerun upgrades after changing bootstrap code.

An exclusive lock prevents concurrent launchers from using the same data directory, including with --fresh. Old stack.json files are removed under that lock before startup; the discovery file is removed again on exit. --fresh clears only a directory previously marked as belonging to this launcher (or an empty directory). With shared OpenSearch it also deletes the exact index prefixes of this stack's current and previously persisted instances, followed by _, including matching aliases, data streams, and templates. Names are enumerated before deletion; no wildcard DELETE is sent. Other prefixes are preserved. Without --fresh, the same prefixes and data are reused. Do not move an active data directory or remove its sibling .gms-lite.lock file. The launcher never removes H2/OpenSearch lock files or reclaims other processes' ports. Startup errors retain data and logs for diagnosis.

Usage-event indices use the mappings from the upgrade jar's source template, without ISM lifecycle settings. Automatic usage-event rollover and retention are unavailable on this minimal node. The embedded distribution includes the matching custom-codecs plugin because GMS selects zstd_no_dict for ordinary indices. Semantic search remains disabled by default; the launcher does not request k-NN.

Verification

./gradlew :metadata-service:gms-lite:check
./gradlew :metadata-ingestion:lintFix
./gradlew :metadata-service:gms-lite:stressTest
# New results directory (previous results are otherwise resumed):
./gradlew :metadata-service:gms-lite:stressTest -PstressDataDir=/tmp/gms-lite-check
# Select cases:
./gradlew :metadata-service:gms-lite:stressTest -PstressTests="persistence parallel stale"
# Or build and verify with a fresh temporary results directory:
metadata-service/gms-lite/bin/verify-stack.sh

launcherSafetyTest runs fast, offline safety tests and is part of check. The module's lint and lintFix Gradle tasks use the repository's pinned Python tool and ruff.toml; check includes lint. A repository pre-commit hook runs the module's lintFix task for Python/config changes.

The opt-in stress suite checks authenticated ingest/read/search, H2 and spool restart persistence, same-PID warm starts, parallel instance isolation, concurrent cold launches of two separate stacks, replacement of dead nodes, timeout without replacement of a paused live node, private-node shutdown, prefix-scoped --fresh, restoration after total or single-core-index loss, SIGTERM during upgrades, closed owned ports, stale discovery recovery, and explicit manager shutdown. It always uses a temporary shared state directory and stops that node in teardown, including after failures. Each case's verdict is appended and flushed immediately to build/stress/results.jsonl. Re-running skips already recorded cases, including failures; use a new --data-dir or remove that results file to rerun them. Results and service logs are retained on failure. Use a port base with two free consecutive ports (default 18090).

tail -f metadata-service/gms-lite/build/stress/results.jsonl
python3 metadata-service/gms-lite/tests/summarize.py \
  metadata-service/gms-lite/build/stress/results.jsonl

Environment contract

All settings below are read by metadata-service/configuration/src/main/resources/application.yaml, except the explicit Java/Spring bindings noted in the final column. Both GMS and upgrades receive this environment; an option applying to only one application is inert in the other. Unlisted inherited or user-supplied settings keep their normal OSS meaning.

Variable	Launcher value	Reader
EBEAN_DATASOURCE_URL	jdbc:h2:file:<data-dir>/<instance>/gms;MODE=MySQL	ebean.url
EBEAN_DATASOURCE_DRIVER	org.h2.Driver	ebean.driver
EBEAN_DATASOURCE_USERNAME	sa	ebean.username
EBEAN_DATASOURCE_PASSWORD	empty	ebean.password
EBEAN_AUTOCREATE	true only for SystemUpdate on a missing H2 database; otherwise false	ebean.autoCreateDdl
DATAHUB_SQL_SETUP_ENABLED	false	SqlSetupCondition Java
DATAHUB_MESSAGING_TRANSPORT	pgqueue	datahub.messaging.transport
DATAHUB_PGQUEUE_ENABLED	true	postgres.pgQueue.enabled
DATAHUB_PGQUEUE_STORE	inmemory	postgres.pgQueue.store
DATAHUB_PGQUEUE_INMEMORY_SPOOL_FILE	<data-dir>/<instance>/queue-spool.json	postgres.pgQueue.inmemorySpoolFile
SCHEMA_REGISTRY_TYPE	INTERNAL	kafka.schemaRegistry.type
KAFKA_SCHEMAREGISTRY_URL	http://localhost:<gms-port>/schema-registry/api	kafka.schemaRegistry.url
MAE_CONSUMER_ENABLED	true	maeConsumer.enabled
MCE_CONSUMER_ENABLED	true	MetadataChangeProposalProcessorCondition and related Java conditions
PE_CONSUMER_ENABLED	true	PlatformEventProcessorCondition Java
DATAHUB_OBJECT_STORAGE_URI	instance-local file: URI	datahub.objectStorage.uri
ELASTICSEARCH_HOST	localhost	elasticsearch.host
ELASTICSEARCH_PORT	embedded HTTP port	elasticsearch.port
ELASTICSEARCH_SHIM_ENGINE_TYPE	OPENSEARCH_3	elasticsearch.clusters.primary.shim.engineType
ELASTICSEARCH_SHIM_AUTO_DETECT	false	elasticsearch.clusters.primary.shim.autoDetectEngine
ELASTICSEARCH_INDEX_DOC_IDS_SCHEMA_FIELD_HASH_ID_ENABLED	true	elasticsearch.entityIndex.v2.docIds.schemaField.hashIdEnabled
INDEX_PREFIX	path hash + instance name	elasticsearch.index.prefix
GRAPH_SERVICE_IMPL	elasticsearch	graphService.type
ELASTICSEARCH_NUM_SHARDS_PER_INDEX	1	elasticsearch.clusters.primary.index.numShards
ELASTICSEARCH_NUM_REPLICAS_PER_INDEX	0	elasticsearch.clusters.primary.index.numReplicas
SKIP_CREATE_USAGE_EVENT_INDICES_STEP	true	CreateUsageEventIndicesStep Java
BOOTSTRAP_SYSTEM_UPDATE_WAIT_FOR_SYSTEM_UPDATE	false	bootstrap.waitForSystemUpdate
METADATA_SERVICE_AUTH_ENABLED	true; false with --no-auth	authentication.enabled
POLICY_CACHE_REFRESH_INTERVAL_SECONDS	10	authorization.defaultAuthorizer.cacheRefreshIntervalSecs
DATAHUB_SYSTEM_CLIENT_ID	__datahub_system	authentication.systemClientId
DATAHUB_SYSTEM_CLIENT_SECRET	random, persisted	authentication.systemClientSecret
DATAHUB_TOKEN_SERVICE_SIGNING_KEY	random, persisted	authentication.tokenService.signingKey
DATAHUB_TOKEN_SERVICE_SALT	random, persisted	authentication.tokenService.salt
ENTITY_REGISTRY_CONFIG_PATH	OSS entity-registry YAML	entityRegistry.path
DATAHUB_GMS_HOST	localhost	datahub.gms.host
DATAHUB_GMS_PORT	instance HTTP port	datahub.gms.port
SERVER_PORT	instance HTTP port	Spring Boot server.port
SERVER_ADDRESS	127.0.0.1	Spring Boot server.address
SPRING_PROFILES_ACTIVE	existing profiles plus upgrade, upgrade JVMs only	Spring Boot profile binding

Neutralised search overrides

These values override both the shell and --env-file. Each YAML property is protected by clearing or pinning both its placeholder names and its canonical environment name. Names are generated from the paths below using Spring Boot's environment mapping: replace dots with underscores, remove dashes, uppercase, and surround list indices with underscores. Camel-case words stay joined (e.g. pathPrefix becomes PATHPREFIX). same means the canonical name is already a listed placeholder.

Empty URI values let ElasticSearchConfiguration.normalizeClusters() synthesize the primary URI from the pinned host/port and leave the secondary unconfigured. elasticsearch.host and .port already map to ELASTICSEARCH_HOST and ELASTICSEARCH_PORT, pinned to localhost and the embedded port. Neither the YAML nor SearchClusterSettings defines clusters.primary.host or .port; nested host/port names cannot override the connection. Boolean switches use false; proxy schemes use http.

YAML key	Placeholder name(s)	Canonical env name	Value
elasticsearch.pathPrefix	ELASTICSEARCH_PATH_PREFIX	ELASTICSEARCH_PATHPREFIX	empty
elasticsearch.useSSL	ELASTICSEARCH_USE_SSL	ELASTICSEARCH_USESSL	false
postgres.pgCron.iam.awsRegion	AWS_REGION	POSTGRES_PGCRON_IAM_AWSREGION	empty
elasticsearch.clusters.primary.uri	ELASTICSEARCH_URI	ELASTICSEARCH_CLUSTERS_PRIMARY_URI	empty
elasticsearch.clusters.primary.username	ELASTICSEARCH_USERNAME	ELASTICSEARCH_CLUSTERS_PRIMARY_USERNAME	empty
elasticsearch.clusters.primary.password	ELASTICSEARCH_PASSWORD	ELASTICSEARCH_CLUSTERS_PRIMARY_PASSWORD	empty
elasticsearch.clusters.primary.region	AWS_REGION	ELASTICSEARCH_CLUSTERS_PRIMARY_REGION	empty
elasticsearch.clusters.primary.opensearchUseAwsIamAuth	OPENSEARCH_USE_AWS_IAM_AUTH	ELASTICSEARCH_CLUSTERS_PRIMARY_OPENSEARCHUSEAWSIAMAUTH	false
elasticsearch.clusters.secondary.uri	ELASTICSEARCH_CLUSTERS_SECONDARY_URI	same	empty
elasticsearch.clusters.secondary.username	ELASTICSEARCH_CLUSTERS_SECONDARY_USERNAME	same	empty
elasticsearch.clusters.secondary.password	ELASTICSEARCH_CLUSTERS_SECONDARY_PASSWORD	same	empty
elasticsearch.clusters.secondary.region	ELASTICSEARCH_CLUSTERS_SECONDARY_REGION	same	empty
elasticsearch.clusters.secondary.opensearchUseAwsIamAuth	ELASTICSEARCH_CLUSTERS_SECONDARY_OPENSEARCH_USE_AWS_IAM_AUTH	ELASTICSEARCH_CLUSTERS_SECONDARY_OPENSEARCHUSEAWSIAMAUTH	false
elasticsearch.clusters.primary.sslContext.protocol	ELASTICSEARCH_SSL_PROTOCOL	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_PROTOCOL	empty
elasticsearch.clusters.primary.sslContext.secureRandomImplementation	ELASTICSEARCH_SSL_SECURE_RANDOM_IMPL	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_SECURERANDOMIMPLEMENTATION	empty
elasticsearch.clusters.primary.sslContext.trustStoreFile	ELASTICSEARCH_SSL_TRUSTSTORE_FILE	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_TRUSTSTOREFILE	empty
elasticsearch.clusters.primary.sslContext.trustStoreType	ELASTICSEARCH_SSL_TRUSTSTORE_TYPE	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_TRUSTSTORETYPE	empty
elasticsearch.clusters.primary.sslContext.trustStorePassword	ELASTICSEARCH_SSL_TRUSTSTORE_PASSWORD	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_TRUSTSTOREPASSWORD	empty
elasticsearch.clusters.primary.sslContext.keyStoreFile	ELASTICSEARCH_SSL_KEYSTORE_FILE	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_KEYSTOREFILE	empty
elasticsearch.clusters.primary.sslContext.keyStoreType	ELASTICSEARCH_SSL_KEYSTORE_TYPE	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_KEYSTORETYPE	empty
elasticsearch.clusters.primary.sslContext.keyStorePassword	ELASTICSEARCH_SSL_KEYSTORE_PASSWORD	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_KEYSTOREPASSWORD	empty
elasticsearch.clusters.primary.sslContext.keyPassword	ELASTICSEARCH_SSL_KEY_PASSWORD	ELASTICSEARCH_CLUSTERS_PRIMARY_SSLCONTEXT_KEYPASSWORD	empty
elasticsearch.clusters.secondary.sslContext.protocol	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_PROTOCOL	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_PROTOCOL	empty
elasticsearch.clusters.secondary.sslContext.secureRandomImplementation	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_SECURE_RANDOM_IMPL	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_SECURERANDOMIMPLEMENTATION	empty
elasticsearch.clusters.secondary.sslContext.trustStoreFile	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_TRUSTSTORE_FILE	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_TRUSTSTOREFILE	empty
elasticsearch.clusters.secondary.sslContext.trustStoreType	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_TRUSTSTORE_TYPE	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_TRUSTSTORETYPE	empty
elasticsearch.clusters.secondary.sslContext.trustStorePassword	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_TRUSTSTORE_PASSWORD	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_TRUSTSTOREPASSWORD	empty
elasticsearch.clusters.secondary.sslContext.keyStoreFile	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_KEYSTORE_FILE	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_KEYSTOREFILE	empty
elasticsearch.clusters.secondary.sslContext.keyStoreType	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_KEYSTORE_TYPE	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_KEYSTORETYPE	empty
elasticsearch.clusters.secondary.sslContext.keyStorePassword	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_KEYSTORE_PASSWORD	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_KEYSTOREPASSWORD	empty
elasticsearch.clusters.secondary.sslContext.keyPassword	ELASTICSEARCH_CLUSTERS_SECONDARY_SSL_KEY_PASSWORD	ELASTICSEARCH_CLUSTERS_SECONDARY_SSLCONTEXT_KEYPASSWORD	empty
elasticsearch.clusters.primary.proxy.host	ELASTICSEARCH_CLUSTERS_PRIMARY_PROXY_HOST, ELASTICSEARCH_PROXY_HOST	same	empty
elasticsearch.clusters.primary.proxy.port	ELASTICSEARCH_CLUSTERS_PRIMARY_PROXY_PORT, ELASTICSEARCH_PROXY_PORT	same	empty
elasticsearch.clusters.primary.proxy.scheme	ELASTICSEARCH_CLUSTERS_PRIMARY_PROXY_SCHEME, ELASTICSEARCH_PROXY_SCHEME	same	http
elasticsearch.clusters.primary.proxy.username	ELASTICSEARCH_CLUSTERS_PRIMARY_PROXY_USERNAME, ELASTICSEARCH_PROXY_USERNAME	same	empty
elasticsearch.clusters.primary.proxy.password	ELASTICSEARCH_CLUSTERS_PRIMARY_PROXY_PASSWORD, ELASTICSEARCH_PROXY_PASSWORD	same	empty
elasticsearch.clusters.secondary.proxy.host	ELASTICSEARCH_CLUSTERS_SECONDARY_PROXY_HOST	same	empty
elasticsearch.clusters.secondary.proxy.port	ELASTICSEARCH_CLUSTERS_SECONDARY_PROXY_PORT	same	empty
elasticsearch.clusters.secondary.proxy.scheme	ELASTICSEARCH_CLUSTERS_SECONDARY_PROXY_SCHEME	same	http
elasticsearch.clusters.secondary.proxy.username	ELASTICSEARCH_CLUSTERS_SECONDARY_PROXY_USERNAME	same	empty
elasticsearch.clusters.secondary.proxy.password	ELASTICSEARCH_CLUSTERS_SECONDARY_PROXY_PASSWORD	same	empty
elasticsearch.clusters.primary.proxy.useSystemProxyProperties	ELASTICSEARCH_CLUSTERS_PRIMARY_PROXY_USE_SYSTEM_PROXY_PROPERTIES	ELASTICSEARCH_CLUSTERS_PRIMARY_PROXY_USESYSTEMPROXYPROPERTIES	false
elasticsearch.clusters.primary.index.prefix	ELASTICSEARCH_CLUSTERS_PRIMARY_INDEX_PREFIX	same	empty
elasticsearch.clusters.secondary.proxy.useSystemProxyProperties	ELASTICSEARCH_CLUSTERS_SECONDARY_PROXY_USE_SYSTEM_PROXY_PROPERTIES	ELASTICSEARCH_CLUSTERS_SECONDARY_PROXY_USESYSTEMPROXYPROPERTIES	false
elasticsearch.clusters.secondary.index.prefix	ELASTICSEARCH_CLUSTERS_SECONDARY_INDEX_PREFIX	same	empty
elasticsearch.componentCluster.searchV2	ELASTICSEARCH_COMPONENT_CLUSTER_SEARCH_V2	ELASTICSEARCH_COMPONENTCLUSTER_SEARCHV2	primary
elasticsearch.componentCluster.searchV3	ELASTICSEARCH_COMPONENT_CLUSTER_SEARCH_V3	ELASTICSEARCH_COMPONENTCLUSTER_SEARCHV3	primary
elasticsearch.componentCluster.semantic	ELASTICSEARCH_COMPONENT_CLUSTER_SEMANTIC	ELASTICSEARCH_COMPONENTCLUSTER_SEMANTIC	primary
elasticsearch.componentCluster.graph	ELASTICSEARCH_COMPONENT_CLUSTER_GRAPH	ELASTICSEARCH_COMPONENTCLUSTER_GRAPH	primary
elasticsearch.componentCluster.timeseries	ELASTICSEARCH_COMPONENT_CLUSTER_TIMESERIES	ELASTICSEARCH_COMPONENTCLUSTER_TIMESERIES	primary
elasticsearch.componentCluster.systemMetadata	ELASTICSEARCH_COMPONENT_CLUSTER_SYSTEM_METADATA	ELASTICSEARCH_COMPONENTCLUSTER_SYSTEMMETADATA	primary
elasticsearch.componentCluster.usage	ELASTICSEARCH_COMPONENT_CLUSTER_USAGE	ELASTICSEARCH_COMPONENTCLUSTER_USAGE	primary
elasticsearch.clusterMigration.graph.targetCluster	ELASTICSEARCH_CLUSTER_MIGRATION_GRAPH_TARGET_CLUSTER	ELASTICSEARCH_CLUSTERMIGRATION_GRAPH_TARGETCLUSTER	empty
elasticsearch.clusterMigration.graph.copy.remote.host	ELASTICSEARCH_CLUSTER_MIGRATION_GRAPH_COPY_REMOTE_HOST	ELASTICSEARCH_CLUSTERMIGRATION_GRAPH_COPY_REMOTE_HOST	empty
elasticsearch.clusterMigration.graph.copy.remote.username	ELASTICSEARCH_CLUSTER_MIGRATION_GRAPH_COPY_REMOTE_USERNAME	ELASTICSEARCH_CLUSTERMIGRATION_GRAPH_COPY_REMOTE_USERNAME	empty
elasticsearch.clusterMigration.graph.copy.remote.password	ELASTICSEARCH_CLUSTER_MIGRATION_GRAPH_COPY_REMOTE_PASSWORD	ELASTICSEARCH_CLUSTERMIGRATION_GRAPH_COPY_REMOTE_PASSWORD	empty
elasticsearch.clusterMigration.timeseries.targetCluster	ELASTICSEARCH_CLUSTER_MIGRATION_TIMESERIES_TARGET_CLUSTER	ELASTICSEARCH_CLUSTERMIGRATION_TIMESERIES_TARGETCLUSTER	empty
elasticsearch.clusterMigration.timeseries.copy.remote.host	ELASTICSEARCH_CLUSTER_MIGRATION_TIMESERIES_COPY_REMOTE_HOST	ELASTICSEARCH_CLUSTERMIGRATION_TIMESERIES_COPY_REMOTE_HOST	empty
elasticsearch.clusterMigration.timeseries.copy.remote.username	ELASTICSEARCH_CLUSTER_MIGRATION_TIMESERIES_COPY_REMOTE_USERNAME	ELASTICSEARCH_CLUSTERMIGRATION_TIMESERIES_COPY_REMOTE_USERNAME	empty
elasticsearch.clusterMigration.timeseries.copy.remote.password	ELASTICSEARCH_CLUSTER_MIGRATION_TIMESERIES_COPY_REMOTE_PASSWORD	ELASTICSEARCH_CLUSTERMIGRATION_TIMESERIES_COPY_REMOTE_PASSWORD	empty
elasticsearch.clusterMigration.systemMetadata.targetCluster	ELASTICSEARCH_CLUSTER_MIGRATION_SYSTEM_METADATA_TARGET_CLUSTER	ELASTICSEARCH_CLUSTERMIGRATION_SYSTEMMETADATA_TARGETCLUSTER	empty
elasticsearch.clusterMigration.systemMetadata.copy.remote.host	ELASTICSEARCH_CLUSTER_MIGRATION_SYSTEM_METADATA_COPY_REMOTE_HOST	ELASTICSEARCH_CLUSTERMIGRATION_SYSTEMMETADATA_COPY_REMOTE_HOST	empty
elasticsearch.clusterMigration.systemMetadata.copy.remote.username	ELASTICSEARCH_CLUSTER_MIGRATION_SYSTEM_METADATA_COPY_REMOTE_USERNAME	ELASTICSEARCH_CLUSTERMIGRATION_SYSTEMMETADATA_COPY_REMOTE_USERNAME	empty
elasticsearch.clusterMigration.systemMetadata.copy.remote.password	ELASTICSEARCH_CLUSTER_MIGRATION_SYSTEM_METADATA_COPY_REMOTE_PASSWORD	ELASTICSEARCH_CLUSTERMIGRATION_SYSTEMMETADATA_COPY_REMOTE_PASSWORD	empty
elasticsearch.clusterMigration.usage.targetCluster	ELASTICSEARCH_CLUSTER_MIGRATION_USAGE_TARGET_CLUSTER	ELASTICSEARCH_CLUSTERMIGRATION_USAGE_TARGETCLUSTER	empty
elasticsearch.clusterMigration.usage.copy.remote.host	ELASTICSEARCH_CLUSTER_MIGRATION_USAGE_COPY_REMOTE_HOST	ELASTICSEARCH_CLUSTERMIGRATION_USAGE_COPY_REMOTE_HOST	empty
elasticsearch.clusterMigration.usage.copy.remote.username	ELASTICSEARCH_CLUSTER_MIGRATION_USAGE_COPY_REMOTE_USERNAME	ELASTICSEARCH_CLUSTERMIGRATION_USAGE_COPY_REMOTE_USERNAME	empty
elasticsearch.clusterMigration.usage.copy.remote.password	ELASTICSEARCH_CLUSTER_MIGRATION_USAGE_COPY_REMOTE_PASSWORD	ELASTICSEARCH_CLUSTERMIGRATION_USAGE_COPY_REMOTE_PASSWORD	empty