Benchmarking results investigation


#1

Hello. I'd like to discuss an issue.

I've got some results after running custom track on elasticsearch index.
Custom track used for benchmarking

Lap Metric Task Value Unit
All Indexing time 0.0715 min
All Indexing throttle time 0 min
All Merge time 0.0864333 min
All Refresh time 0.132483 min
All Flush time 0.00181667 min
All Merge throttle time 0 min
All Total Young Gen GC 0.062 s
All Total Old Gen GC 0 s
All Store size 0.976847 GB
All Translog size 0.0157164 GB
All Heap used for segments 2.97031 MB
All Heap used for doc values 0.553997 MB
All Heap used for terms 2.17584 MB
All Heap used for norms 0.0727539 MB
All Heap used for points 0.0613871 MB
All Heap used for stored fields 0.106339 MB
All Segment count 109
All 50th percentile latency test generic request time 2.09754 ms
All 90th percentile latency test generic request time 3.99158 ms
All 99th percentile latency test generic request time 8.00824 ms
All 99.9th percentile latency test generic request time 26.267 ms
All 100th percentile latency test generic request time 43.0159 ms
All 50th percentile service time test generic request time 1.96887 ms
All 90th percentile service time test generic request time 3.84714 ms
All 99th percentile service time test generic request time 6.25481 ms
All 99.9th percentile service time test generic request time 10.8736 ms
All 100th percentile service time test generic request time 12.8791 ms
All error rate test generic request time 100 %

2018-05-24 12:15:49,565 PID:9160 rally.reporting WARNING No throughput metrics available for [test generic request time]. Likely cause: Error rate is 100.0%. Please check the logs.
2018-05-24 12:15:49,566 PID:9160 rally.metrics INFO Closing metrics store.
2018-05-24 12:15:49,567 PID:9152 rally.racecontrol INFO Benchmark has finished successfully.
2018-05-24 12:15:49,567 PID:9152 rally.racecontrol INFO Telling benchmark actor to exit.
2018-05-24 12:15:49,568 PID:9160 rally.racecontrol INFO BenchmarkActor received unknown message [ActorExitRequest] (ignoring).

However as you can see I've got 100% of error rate.

Looking into the logs there is next point that could be connected with error rate metric beeing 100%.
Every single query results with a HTTP 400 Bad Request Response

2018-05-24 12:15:48,917 PID:9222 elasticsearch WARNING GET http://172.19.0.3:9200/index_name_20180516105114/_search?request_cache=false [status:400 request:0.001s]

On the other hand running same query via curl from my local machine responds with a traditional 200 response and a document returned by elasticsearch.

Do you have any thoughts about possible issue?

Thank you.


(Daniel Mitterdorfer) #2

Hi,

I had a look at your track and you specify _source inside query but this should be specified on top-level instead. So the correct query body is:

{
  "_source": [
    "property_type_id",
    "park_code",
    "platform_options.bvdir",
    "property_type_id",
    "parameters",
    "location",
    "rating",
    "stars",
    "number_of_survays",
    "labels",
    "park_code",
    "coordinates",
    "id",
    "availabilities_with_min_prices",
    "availabilities_with_max_prices"
  ],
  "from": 0,
  "size": 10,
  "sort": [
    {
      "_score": "desc"
    },
    {
      "sorting.global_points_2": "desc"
    }
  ],
  "query": {
    "bool": {
      "filter": [
        {
          "term": {
            "platform": "bvdir"
          }
        },
        {
          "term": {
            "demo": false
          }
        }
      ]
    }
  }
}

You can also try to run Rally with --on-error="abort" and see more error details.

Daniel


#3

Thank you Daniel.
Finally it works for me. And thats great. I'm happy!
Have inspiration to delve into esrally docs to experiment more widely with the tool.
Cheers!


(Daniel Mitterdorfer) #4

Hi Adam,

glad to hear that it's resolved now. :slight_smile:

Daniel


(system) #5

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.