WARNING: Version 2.4 has passed its EOL date.

This documentation is no longer being maintained and may be removed. If you are running this version, we strongly advise you to upgrade. For the latest information, see the current release documentation.

« Mapping and Types Performance considerations »
Elastic Docs ›Elasticsearch for Apache Hadoop [2.4] ›Elasticsearch for Apache Hadoop

Hadoop Metrics

edit

Hadoop Metrics

edit

The Hadoop system records a set of metric counters for each job that it runs. elasticsearch-hadoop extends on that and provides metrics about its activity for each job run by leveraging the Hadoop Counters infrastructure. During each run, elasticsearch-hadoop sends statistics from each task instance, as it is running, which get aggregated by the Map/Reduce infrastructure and are available through the standard Hadoop APIs.

elasticsearch-hadoop provides the following counters, available under org.elasticsearch.hadoop.mr.Counter enum:

Table 8. Available counters

Counter name Purpose

Data focused

BYTES_SENT

Total number of data/communication bytes sent over the network to Elasticsearch

BYTES_ACCEPTED

Data/Documents accepted by Elasticsearch in bytes

BYTES_RETRIED

Data/Documents rejected by Elasticsearch in bytes

BYTES_RECEIVED

Data/Documents received from Elasticsearch in bytes

Document focused

DOCS_SENT

Number of docs sent over the network to Elasticsearch

DOCS_ACCEPTED

Number of documents sent and accepted by Elasticsearch

DOCS_RETRIED

Number of documents sent but rejected by Elasticsearch

DOCS_RECEIVED

Number of documents received from Elasticsearch

Network focused

BULK_TOTAL

Number of bulk requests made to Elasticsearch

BULK_RETRIES

Number of bulk retries (caused by document rejections)

SCROLL_TOTAL

Number of scroll pulled from Elasticsearch

NODE_RETRIES

Number of node fall backs (caused by network errors)

NET_RETRIES

Number of network retries (caused by network errors)

Time focused

NET_TOTAL_TIME_MS

Overall time (in ms) spent over the network

BULK_TOTAL_TIME_MS

Time (in ms) spent over the network by the bulk requests

BULK_RETRIES_TOTAL_TIME_MS

Time (in ms) spent over the network retrying bulk requests

SCROLL_TOTAL_TIME_MS

Time (in ms) spent over the network reading the scroll requests

One can use the counters programatically, depending on the API used, through mapred or mapreduce. Whatever the choice, elasticsearch-hadoop performs automatic reports without any user intervention. In fact, when using elasticsearch-hadoop one will see the stats reported at the end of the job run, for example:

13:55:08,100  INFO main mapreduce.Job - Job job_local127738678_0013 completed successfully
13:55:08,101  INFO main mapreduce.Job - Counters: 35
...
Elasticsearch Hadoop Counters
    Bulk Retries=0
    Bulk Retries Total Time(ms)=0
    Bulk Total=20
    Bulk Total Time(ms)=518
    Bytes Accepted=159129
    Bytes Sent=159129
    Bytes Received=79921
    Bytes Retried=0
    Documents Accepted=993
    Documents Sent=993
    Documents Received=0
    Documents Retried=0
    Network Retries=0
    Network Total Time(ms)=937
    Node Retries=0
    Scroll Total=0
    Scroll Total Time(ms)=0
« Mapping and Types Performance considerations »