[
https://issues.apache.org/jira/browse/IGNITE-12189?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16948677#comment-16948677
]
Andrey Mashenkov commented on IGNITE-12189:
-------------------------------------------
Thanks.
I've just found you apply 'limit' on reduce side inside CacheQueryFutureAdapter.
Assume, you have few nodes, small pageSize (10 entries per page) and large
enough 'limit' = 1000.
So, you'll get about 100 pages from every node and about 1000 in total, but
most likely you want only first 100 pages.
For now query initiator node sends 1 GridCacheQueryRequest and can get multiple
GridCacheQueryResponse-s per remote.
TextQuery processing finishes when all remotes send GridCacheQueryResponse with
'finished' flag.
Seems, TextQuery processing should be reworked like it was done for SQL
queries.
Ignite has GridQueryNextPageRequest \ Response classes for SQL result
processing.
Similar processing should be implemented for TextQueries.
This looks like a big change which allow us to add correct sorting and apply
limit correctly on reduce .
Let's create a separate ticket for this and link it to current one, then I'll
merge PR to master.
> Implement correct limit for TextQuery
> -------------------------------------
>
> Key: IGNITE-12189
> URL: https://issues.apache.org/jira/browse/IGNITE-12189
> Project: Ignite
> Issue Type: Improvement
> Components: general
> Reporter: Yuriy Shuliha
> Assignee: Yuriy Shuliha
> Priority: Major
> Fix For: 2.8
>
> Time Spent: 2h 40m
> Remaining Estimate: 0h
>
> PROBLEM
> For now each server-node returns all response records to the client-node and
> it may contain ~thousands, ~hundred thousands records.
> Event if we need only first 10-100. Again, all the results are added to
> queue in _*GridCacheQueryFutureAdapter*_ in arbitrary order by pages.
> There are no any means to deliver deterministic result.
> SOLUTION
> Implement _*limit*_ as parameter for _*TextQuery*_ and
> _*GridCacheQueryRequest*_
> It should be passed as limit parameter in Lucene's
> _*IndexSearcher.search()*_ in _*GridLuceneIndex*_.
> For distributed queries _*limit*_ will also trim response queue when merging
> results.
> Type: long
> Special value: : 0 -> No limit (Integer.MAX_VALUE);
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
