Percolator Performance (ES 5.6)

Hi,

Since moving from 2.x to 5.6 we have seen a drastic increase in overall query time, and one the biggest culprits is percolation.

After some research, I've been looking at the "query.extraction_result" : "failed" count to see which queries aren't being extracted along with the count.

A couple questions:

Why would this seemingly simple query not be extracted? Is there an API to determine why a given query wouldn't be extracted?

"query": {
  "bool": {
    "filter": [
      {
        "range": {
          "some_field": {
            "gte": 100
          }
        }
      }
    ]
  }
}

Also how many un-extracted queries is too many? I realize this is almost certainly going to be dependent on other factors, but I just want to get sense of if it's worth pulling on this thread.

Is there anything else that could contribute to excessive percolator query times?

Thanks!

1 Like