Hello:
I wonder - we have the following software:
| Search | 6.0.1 |
| Cassandra | 3.11.3.5113 |
| DSE | 5.1.13 |
Some of our queries are non-performant. Is setting timeAllowed will help us to eliminate this problem? We deployed the code with such a setting - however we see no effect, nor warnings from DSESolr.
Here's more info on that:
https://docs.datastax.com/en/dse/5.1/dse-dev/datastax_enterprise/search/limitQueriesByTime.html
OSS Solr's TimeLimitingCollector, which is used by DSE Search, suffers from two problems. The first is that for many queries that are likely to take an inordinately long time, like pivot facet queries, it has essentially no effect, because it is evaluated during the Lucene collection phase. The second is that even there, it carries with it a significant performance penalty for queries that won't hit the time limit in the first place.
Having said that, we are currently working on a timeAllowed mechanism that addresses both these issues, and that will land in an upcoming 6.0.x patch release.
@maedhroz - Would you please refer Jira tickets for aforementioned 'upcoming 6.0.x patch release' that deals with Lucene's TimeLimitingCollector class and 'timeAllowed' parameter? I would greatly appreciate that.
3 People are following this question.
