Elastic Data Ingesting

Introduction

With ingesting pipelines, incoming events or events already present on the server can be easily and transparently modified and extended. They are comparable to Logstash pipelines, but do not have the same functionality. However, because they are executed in the Elasticsearch process, ingest pipelines are much easier to use.

In the context of service.monitor, there are currently the following pipelines:

  • Calculation of extended time and date values

  • Detailed examination of Esri ArcGIS log data

  • Extract ArcGIS service information from Monitoring events

The pipeline files are located in the delivery under elasticsearch/ingest.

Use Cases

Temporal Adjustments

The pipeline ct-monitor-temporal allows the calculation of the following additional time values:

  • day of week

  • name of day of week

  • hour of day

  • day of year

  • number of month

This can be used to create diagrams that, for example, visualise or aggregate measured values on a daily or weekly basis.

By calculating these values, a defined time zone is used (here: 'Europe/Paris'), this must be adjusted if necessary.

In-depth ArcGIS analysis

The pipeline ct-monitor-arcgis-parse-servicename allows the extraction of service information from ArcGIS Server messages. In most cases the service concerned is stored in a separate attribute of the ArcGIS message, but in some cases it is only present in the message text of the log line. The pipeline is able to recognise service names via Grok patterns and transfer them into the standard field of the event.

This pipeline helps to find service timeouts, for example. ("Error performing query operation Error handling service request :Processing request took longer than the usage timeout for service")

Extract ArcGIS service information and host name from Monitoring events

The ct-monitor-arcgis-serviceinfo pipeline makes it possible to extract ArcGIS service information from Monitoring events.

Fields added by the ct-monitor-arcgis-serviceinfo pipeline
[...]
"service" : {
  [...]
  "host" : "services.conterra.de",
  "ags": {
     "url-path" : "/arcgis/rest/services/mapapps/stoerung_relates/MapServer",
     "arcgis-service-name" : "mapapps",
     "arcgis-folder" : "stoerung_relates",
     "arcgis-service-and-folder" : "mapapps/stoerung_relates"
  },
  [...]
}

Usage

Running pipelines for inventory data

Applying a pipeline during reindexation
POST _reindex
{
  "source": {
    "index": "source_index"
  },
  "dest": {
    "index": "target_index",
    "pipeline": "pipeline-name"
  }
}

The _update_by_query feature can also be used to update old data. In the example below, a query is also used to apply the pipeline only to certain events.

Applying a pipeline "in place"
POST <index>/_update_by_query?pipeline=pipeline-name
{
  "query": {
    "match": {
      "_id": "9_BGMnoBoGFrpZgJVWVE"
    }
  }
}

Execute pipelines for all incoming events

By setting pipeline settings, a pipeline can be set as the default pipeline. Pipeline für Index konfigurieren

Set the same value on the associated index template to enable the pipeline on all subsequent indexes as well.