Launch a job
Launch a job
Create a new job run using the specified job configuration. The job’s id must be in the request body.
POST
Launch a job
Body
application/json
The name of the job configuration. This is distinct from the job run ID (found in job status responses), which identifies a specific job run.
Example:
"api-test-app_click_signals_aggregation"
The configuration for this Spark job. The available keys depend on the job type. You can use the /spark/schema endpoint to fetch the configuration schemas for all Spark job types.
The Spark job type.
Available options:
aggregation, als_recommender, argo-classification, argo-data-augmentation, argo-delete-model, argo-delete-ray-model, argo-deploy-model, argo-deploy-ray-model, argo-item-recommender-content, argo-item-recommender-user, argo-milvus-create-collections, argo-milvus-create-indexes, argo-milvus-delete-collections, argo-milvus-delete-indexes, argo-qna-coldstart, argo-qna-evaluate, argo-qna-supervised, argo-upload-model, build-training, cluster_labeling, custom_python_job, custom_spark_scala_job, doc_clustering, experiment_sql, ground_truth, headTailAnalysis, logistic_regression_classifier_trainer, outlier_detection, parallel-bulk-loader, query_similarity, random_forests_classifier, ranking_metrics, script, similar_queries, sip, sql_template, synonymDetection, tokenPhraseSpellCorrection, transfer, trending-recommender, word2vec Response
200 - application/hal+json
OK
Available options:
unknown, idle, starting, running, finishing, cancelling, finished, cancelled, error, skipped Example:
"finished"
The job run ID. This is distinct from id which is the name of the job configuration.
Example:
"hhzitrwancwv"
Example:
{
"type": "aggregation",
"id": "api-test-app_click_signals_aggregation",
"inputCollection": "api-test-app_signals",
"sourceCatchup": true,
"sourceRemove": false,
"sql": "WITH sigs_with_filters AS ( SELECT c.query as query, c.doc_id, q.filters_s as filters, c.type, c.ref_time, coalesce(c.count_i,1) as count_i, c.timestamp_tdt, greatest(coalesce(c.weight_d,0.1),0.0) as weight_d FROM api-test-app_signals c LEFT JOIN (SELECT id, filters_s FROM api-test-app_signals WHERE type='response') q ON q.id = c.fusion_query_id WHERE c.type IN ('click','cart','purchase') AND c.timestamp_tdt >= c.catchup_timestamp_tdt ), signal_type_groups AS ( SELECT SUM(count_i) AS typed_aggr_count_i, query, doc_id, type, filters, time_decay(count_i, timestamp_tdt, \"30 days\", ref_time, weight_d) AS typed_weight_d FROM sigs_with_filters GROUP BY doc_id, query, filters, type ) SELECT concat_ws('|', query, doc_id, filters) as id, SUM(typed_aggr_count_i) AS aggr_count_i, query AS query_s, query AS query_t, doc_id AS doc_id_s, filters AS filters_s, SPLIT(filters, ' \\\\$ ') AS filters_ss, weighted_sum(typed_weight_d, type, 'click:1.0,cart:10.0,purchase:25.0') AS weight_d FROM signal_type_groups GROUP BY query, doc_id, filters",
"rollupSql": "SELECT concat_ws('|', query_s, doc_id_s, filters_s) as id,\n query_s,\n query_s as query_t,\n doc_id_s,\n filters_s,\n first(aggr_type_s) AS aggr_type_s,\n SPLIT(filters_s, ' \\\\$ ') AS filters_ss,\n SUM(weight_d) AS weight_d,\n SUM(aggr_count_i) AS aggr_count_i\n FROM api-test-app_signals_aggr\n GROUP BY query_s, doc_id_s, filters_s",
"referenceTime": "2025-10-17T18:56:14.660Z",
"skipCheckEnabled": true,
"readOptions": [{ "key": "splits_per_shard", "value": "4" }],
"skipJobIfSignalsEmpty": true,
"parameters": [
{
"key": "signalTypeWeights",
"value": "click:1.0,cart:10.0,purchase:25.0"
},
{
"key": "signalTypes",
"value": "_regex/signalTypeWeights/([\\w\\-\\.]*):([\\d\\.\\-]*)(,|$)/'$1'$3/g"
}
],
"selectQuery": "*:*",
"outputCollection": "api-test-app_signals_aggr",
"useNaturalKey": true,
"optimizeSegments": 0,
"dataFormat": "solr",
"sparkSQL": "SELECT * from spark_input",
"sparkPartitions": 200
}Example:
"driver-api-test-app-click-signal-sjnamxwxsyhq"
Example:
{
"jobConfigId": "api-test-app_click_signals_aggregation",
"jobRunId": "sjnamxwxsyhq",
"aggrClass": "SQL",
"query": "WITH sigs_with_filters AS (SELECT c.query as query, c.doc_id, q.filters_s as filters, c.type, c.ref_time, coalesce(c.count_i,1) as count_i, c.timestamp_tdt, greatest(coalesce(c.weight_d,0.1),0.0) as weight_d FROM api_test_app_signals c LEFT JOIN (SELECT id, filters_s FROM api_test_app_signals WHERE type='response') q ON q.id = c.fusion_query_id WHERE c.type IN ('click','cart','purchase') AND c.timestamp_tdt >= c.catchup_timestamp_tdt ), signal_type_groups AS (SELECT SUM(count_i) AS typed_aggr_count_i, query, doc_id, type, filters, time_decay(count_i, timestamp_tdt, \"30 days\", ref_time, weight_d) AS typed_weight_d FROM sigs_with_filters GROUP BY doc_id, query, filters, type) SELECT concat_ws('|', query, doc_id, filters) as id, SUM(typed_aggr_count_i) AS aggr_count_i, query AS query_s, query AS query_t, doc_id AS doc_id_s, filters AS filters_s, SPLIT(filters, ' \\\\$ ') AS filters_ss, weighted_sum(typed_weight_d, type, 'click:1.0,cart:10.0,purchase:25.0') AS weight_d FROM signal_type_groups GROUP BY query, doc_id, filters",
"state": "finished",
"aggregated": 0,
"applicationId": "spark-0424dda9215441c0b232210af1d7cb67",
"podId": "driver-api-test-app-click-signal-sjnamxwxsyhq",
"aggr_type_s": "click@doc_id,filters,query"
}Example:
"2025-10-16T19:49:15.106Z"
Example:
"2025-10-16T19:49:42.276Z"
Example:
27170