Get anomaly detection job statistics API

Retrieves usage information for anomaly detection jobs.

Request

GET _ml/anomaly_detectors/<job_id>/_stats

GET _ml/anomaly_detectors/<job_id>,<job_id>/_stats

GET _ml/anomaly_detectors/_stats

GET _ml/anomaly_detectors/_all/_stats

Prerequisites

Description

You can get statistics for multiple anomaly detection jobs in a single API request by using a group name, a comma-separated list of jobs, or a wildcard expression. You can get statistics for all anomaly detection jobs by using _all, by specifying * as the <job_id>, or by omitting the <job_id>.

This API returns a maximum of 10,000 jobs.

Path parameters

<job_id>

(Optional, string) Identifier for the anomaly detection job. It can be a job identifier, a group name, or a wildcard expression. If you do not specify one of these options, the API returns information for all anomaly detection jobs.

Query parameters

allow_no_jobs

(Optional, boolean) Specifies what to do when the request:

  • Contains wildcard expressions and there are no jobs that match.
  • Contains the _all string or no identifiers and there are no matches.
  • Contains wildcard expressions and there are only partial matches.

The default value is true, which returns an empty jobs array when there are no matches and the subset of results when there are partial matches. If this parameter is false, the request returns a 404 status code when there are no matches or only partial matches.

Response body

The API returns the following information about the operational progress of a job:

assignment_explanation

(string) For open anomaly detection jobs only, contains messages relating to the selection of a node to run the job.

data_counts

(object) An object that describes the quantity of input to the job and any related error counts. The data_count values are cumulative for the lifetime of a job. If a model snapshot is reverted or old results are deleted, the job counts are not reset.

Properties of data_counts

  • bucket_count

    (long) The number of bucket results produced by the job.

    earliest_record_timestamp

    (date) The timestamp of the earliest chronologically input document.

    empty_bucket_count

    (long) The number of buckets which did not contain any data. If your data contains many empty buckets, consider increasing your bucket_span or using functions that are tolerant to gaps in data such as mean, non_null_sum or non_zero_count.

    input_bytes

    (long) The number of bytes of input data posted to the anomaly detection job.

    input_field_count

    (long) The total number of fields in input documents posted to the anomaly detection job. This count includes fields that are not used in the analysis. However, be aware that if you are using a datafeed, it extracts only the required fields from the documents it retrieves before posting them to the job.

    input_record_count

    (long) The number of input documents posted to the anomaly detection job.

    invalid_date_count

    (long) The number of input documents with either a missing date field or a date that could not be parsed.

    job_id

    (string) Identifier for the anomaly detection job.

    last_data_time

    (date) The timestamp at which data was last analyzed, according to server time.

    latest_empty_bucket_timestamp

    (date) The timestamp of the last bucket that did not contain any data.

    latest_record_timestamp

    (date) The timestamp of the latest chronologically input document.

    latest_sparse_bucket_timestamp

    (date) The timestamp of the last bucket that was considered sparse.

    missing_field_count

    (long) The number of input documents that are missing a field that the anomaly detection job is configured to analyze. Input documents with missing fields are still processed because it is possible that not all fields are missing.

    If you are using datafeeds or posting data to the job in JSON format, a high missing_field_count is often not an indication of data issues. It is not necessarily a cause for concern.

    The value of processed_record_count includes this count.

    out_of_order_timestamp_count

    (long) The number of input documents that are out of time sequence and outside of the latency window. This information is applicable only when you provide data to the anomaly detection job by using the post data API. These out of order documents are discarded, since jobs require time series data to be in ascending chronological order.

    processed_field_count

    The total number of fields in all the documents that have been processed by the anomaly detection job. Only fields that are specified in the detector configuration object contribute to this count. The timestamp is not included in this count.

    processed_record_count

    (long) The number of input documents that have been processed by the anomaly detection job. This value includes documents with missing fields, since they are nonetheless analyzed. If you use datafeeds and have aggregations in your search query, the processed_record_count is the number of aggregation results processed, not the number of Elasticsearch documents.

    sparse_bucket_count

    (long) The number of buckets that contained few data points compared to the expected number of data points. If your data contains many sparse buckets, consider using a longer bucket_span.

deleting

(boolean) Indicates that the process of deleting the job is in progress but not yet completed. It is only reported when true.

forecasts_stats

(object) An object that provides statistical information about forecasts belonging to this job. Some statistics are omitted if no forecasts have been made.

Unless there is at least one forecast, memory_bytes, records, processing_time_ms and status properties are omitted.

Properties of forecasts_stats

  • forecasted_jobs

    (long) A value of 0 indicates that forecasts do not exist for this job. A value of 1 indicates that at least one forecast exists.

    memory_bytes

    (object) The avg, min, max and total memory usage in bytes for forecasts related to this job. If there are no forecasts, this property is omitted.

    processing_time_ms

    (object) The avg, min, max and total runtime in milliseconds for forecasts related to this job. If there are no forecasts, this property is omitted.

    records

    (object) The avg, min, max and total number of model_forecast documents written for forecasts related to this job. If there are no forecasts, this property is omitted.

    status

    (object) The count of forecasts by their status. For example: {“finished” : 2, “started” : 1}. If there are no forecasts, this property is omitted.

    total

    (long) The number of individual forecasts currently available for the job. A value of 1 or more indicates that forecasts exist.

job_id

(string) Identifier for the anomaly detection job.

model_size_stats

(object) An object that provides information about the size and contents of the model.

Properties of model_size_stats

  • bucket_allocation_failures_count

    (long) The number of buckets for which new entities in incoming data were not processed due to insufficient model memory. This situation is also signified by a hard_limit: memory_status property value.

    categorized_doc_count

    (long) The number of documents that have had a field categorized.

    categorization_status

    (string) The status of categorization for the job. Contains one of the following values:

    • ok: Categorization is performing acceptably well (or not being used at all).
    • warn: Categorization is detecting a distribution of categories that suggests the input data is inappropriate for categorization. Problems could be that there is only one category, more than 90% of categories are rare, the number of categories is greater than 50% of the number of categorized documents, there are no frequently matched categories, or more than 50% of categories are dead.

    dead_category_count

    (long) The number of categories created by categorization that will never be assigned again because another category’s definition makes it a superset of the dead category. (Dead categories are a side effect of the way categorization has no prior training.)

    failed_category_count

    (long) The number of times that categorization wanted to create a new category but couldn’t because the job had hit its model_memory_limit. This count does not track which specific categories failed to be created. Therefore you cannot use this value to determine the number of unique categories that were missed.

    frequent_category_count

    (long) The number of categories that match more than 1% of categorized documents.

    job_id

    (string) Identifier for the anomaly detection job.

    log_time

    (date) The timestamp of the model_size_stats according to server time.

    memory_status

    (string) The status of the mathematical models, which can have one of the following values:

    • ok: The models stayed below the configured value.
    • soft_limit: The models used more than 60% of the configured memory limit and older unused models will be pruned to free up space.
    • hard_limit: The models used more space than the configured memory limit. As a result, not all incoming data was processed.

    model_bytes

    (long) The number of bytes of memory used by the models. This is the maximum value since the last time the model was persisted. If the job is closed, this value indicates the latest size.

    model_bytes_exceeded

    (long) The number of bytes over the high limit for memory usage at the last allocation failure.

    model_bytes_memory_limit

    (long) The upper limit for model memory usage, checked on increasing values.

    peak_model_bytes

    (long) The peak number of bytes of memory ever used by the models.

    rare_category_count

    (long) The number of categories that match just one categorized document.

    result_type

    (string) For internal use. The type of result.

    total_by_field_count

    (long) The number of by field values that were analyzed by the models. This value is cumulative for all detectors in the job.

    total_category_count

    (long) The number of categories created by categorization.

    total_over_field_count

    (long) The number of over field values that were analyzed by the models. This value is cumulative for all detectors in the job.

    total_partition_field_count

    (long) The number of partition field values that were analyzed by the models. This value is cumulative for all detectors in the job.

    timestamp

    (date) The timestamp of the last record when the model stats were gathered.

node

(object) Contains properties for the node that runs the job. This information is available only for open jobs.

Properties of node

  • attributes

    (object) Lists node attributes such as ml.machine_memory or ml.max_open_jobs settings.

    ephemeral_id

    (string) The ephemeral ID of the node.

    id

    (string) The unique identifier of the node.

    name

    (string) The node name.

    transport_address

    (string) The host and port where transport HTTP connections are accepted.

open_time

(string) For open jobs only, the elapsed time for which the job has been open.

state

(string) The status of the anomaly detection job, which can be one of the following values:

  • closed: The job finished successfully with its model state persisted. The job must be opened before it can accept further data.
  • closing: The job close action is in progress and has not yet completed. A closing job cannot accept further data.
  • failed: The job did not finish successfully due to an error. This situation can occur due to invalid input data, a fatal error occurring during the analysis, or an external interaction such as the process being killed by the Linux out of memory (OOM) killer. If the job had irrevocably failed, it must be force closed and then deleted. If the datafeed can be corrected, the job can be closed and then re-opened.
  • opened: The job is available to receive and process data.
  • opening: The job open action is in progress and has not yet completed.

timing_stats

(object) An object that provides statistical information about timing aspect of this job.

Properties of timing_stats

  • average_bucket_processing_time_ms

    (double) Average of all bucket processing times in milliseconds.

    bucket_count

    (long) The number of buckets processed.

    exponential_average_bucket_processing_time_ms

    (double) Exponential moving average of all bucket processing times, in milliseconds.

    exponential_average_bucket_processing_time_per_hour_ms

    (double) Exponentially-weighted moving average of bucket processing times calculated in a 1 hour time window, in milliseconds.

    job_id

    (string) Identifier for the anomaly detection job.

    maximum_bucket_processing_time_ms

    (double) Maximum among all bucket processing times, in milliseconds.

    minimum_bucket_processing_time_ms

    (double) Minimum among all bucket processing times, in milliseconds.

    total_bucket_processing_time_ms

    (double) Sum of all bucket processing times, in milliseconds.

Response codes

404 (Missing resources)

If allow_no_jobs is false, this code indicates that there are no resources that match the request or only partial matches for the request.

Examples

  1. GET _ml/anomaly_detectors/low_request_rate/_stats

The API returns the following results:

  1. {
  2. "count" : 1,
  3. "jobs" : [
  4. {
  5. "job_id" : "low_request_rate",
  6. "data_counts" : {
  7. "job_id" : "low_request_rate",
  8. "processed_record_count" : 1216,
  9. "processed_field_count" : 1216,
  10. "input_bytes" : 51678,
  11. "input_field_count" : 1216,
  12. "invalid_date_count" : 0,
  13. "missing_field_count" : 0,
  14. "out_of_order_timestamp_count" : 0,
  15. "empty_bucket_count" : 242,
  16. "sparse_bucket_count" : 0,
  17. "bucket_count" : 1457,
  18. "earliest_record_timestamp" : 1575172659612,
  19. "latest_record_timestamp" : 1580417369440,
  20. "last_data_time" : 1576017595046,
  21. "latest_empty_bucket_timestamp" : 1580356800000,
  22. "input_record_count" : 1216
  23. },
  24. "model_size_stats" : {
  25. "job_id" : "low_request_rate",
  26. "result_type" : "model_size_stats",
  27. "model_bytes" : 41480,
  28. "model_bytes_exceeded" : 0,
  29. "model_bytes_memory_limit" : 10485760,
  30. "total_by_field_count" : 3,
  31. "total_over_field_count" : 0,
  32. "total_partition_field_count" : 2,
  33. "bucket_allocation_failures_count" : 0,
  34. "memory_status" : "ok",
  35. "categorized_doc_count" : 0,
  36. "total_category_count" : 0,
  37. "frequent_category_count" : 0,
  38. "rare_category_count" : 0,
  39. "dead_category_count" : 0,
  40. "failed_category_count" : 0,
  41. "categorization_status" : "ok",
  42. "log_time" : 1576017596000,
  43. "timestamp" : 1580410800000
  44. },
  45. "forecasts_stats" : {
  46. "total" : 1,
  47. "forecasted_jobs" : 1,
  48. "memory_bytes" : {
  49. "total" : 9179.0,
  50. "min" : 9179.0,
  51. "avg" : 9179.0,
  52. "max" : 9179.0
  53. },
  54. "records" : {
  55. "total" : 168.0,
  56. "min" : 168.0,
  57. "avg" : 168.0,
  58. "max" : 168.0
  59. },
  60. "processing_time_ms" : {
  61. "total" : 40.0,
  62. "min" : 40.0,
  63. "avg" : 40.0,
  64. "max" : 40.0
  65. },
  66. "status" : {
  67. "finished" : 1
  68. }
  69. },
  70. "state" : "opened",
  71. "node" : {
  72. "id" : "7bmMXyWCRs-TuPfGJJ_yMw",
  73. "name" : "node-0",
  74. "ephemeral_id" : "hoXMLZB0RWKfR9UPPUCxXX",
  75. "transport_address" : "127.0.0.1:9300",
  76. "attributes" : {
  77. "ml.machine_memory" : "17179869184",
  78. "xpack.installed" : "true",
  79. "ml.max_open_jobs" : "20"
  80. }
  81. },
  82. "assignment_explanation" : "",
  83. "open_time" : "13s",
  84. "timing_stats" : {
  85. "job_id" : "low_request_rate",
  86. "bucket_count" : 1457,
  87. "total_bucket_processing_time_ms" : 1094.000000000001,
  88. "minimum_bucket_processing_time_ms" : 0.0,
  89. "maximum_bucket_processing_time_ms" : 48.0,
  90. "average_bucket_processing_time_ms" : 0.75085792724777,
  91. "exponential_average_bucket_processing_time_ms" : 0.5571716855800993,
  92. "exponential_average_bucket_processing_time_per_hour_ms" : 15.0
  93. }
  94. }
  95. ]
  96. }