SOLR-13350, SOLR-17298: multi-threaded search: don't (yet) support query limits

[cpoerschke]

Code change only here. Separately we still need to document that the multiThreaded=true parameter is not yet supported for queries with query limits amongst other things.

https://issues.apache.org/jira/browse/SOLR-13350
https://issues.apache.org/jira/browse/SOLR-17298

[cpoerschke]

https://github.com/apache/solr/actions/runs/10079922542/job/27868413918?pr=2590 CI failure does sound related, will investigate.

> Task :solr:core:wipeTaskTemp
ERROR: The following test(s) have failed:
  - org.apache.solr.ltr.TestLTRQParserPlugin.ltr_expensiveFeatureRescoringAndPartialResultsNotTolerated_shouldRaiseException (:solr:modules:ltr)
    Test history: https://ge.apache.org/scans/tests?search.rootProjectNames=solr-root&tests.container=org.apache.solr.ltr.TestLTRQParserPlugin&tests.test=ltr_expensiveFeatureRescoringAndPartialResultsNotTolerated_shouldRaiseException http://fucit.org/solr-jenkins-reports/history-trend-of-recent-failures.html#series/org.apache.solr.ltr.TestLTRQParserPlugin.ltr_expensiveFeatureRescoringAndPartialResultsNotTolerated_shouldRaiseException
    Test output: /tmp/src/solr/solr/modules/ltr/build/test-results/test/outputs/OUTPUT-org.apache.solr.ltr.TestLTRQParserPlugin.txt
    Reproduce with: ./gradlew :solr:modules:ltr:test --tests "org.apache.solr.ltr.TestLTRQParserPlugin.ltr_expensiveFeatureRescoringAndPartialResultsNotTolerated_shouldRaiseException" -Ptests.jvms=96 "-Ptests.jvmargs=-XX:TieredStopAtLevel=1 -XX:+UseParallelGC -XX:ActiveProcessorCount=1 -XX:ReservedCodeCacheSize=120m" -Ptests.seed=9817284D4195A5AA -Ptests.timeoutSuite=600000! -Ptests.file.encoding=ISO-8859-1

2> 13276 ERROR (TEST-TestLTRQParserPlugin.ltr_expensiveFeatureRescoringAndPartialResultsNotTolerated_shouldRaiseException-seed#[9817284D4195A5AA]) [n: c: s: r: x: t:] o.a.s.u.RestTestBase query failed JSON validation. error: Path not found: /error/msg
  2>  expected: /error/msg=='org.apache.solr.search.QueryLimitsExceededException: Limits exceeded! (Learning To Rank rescoring - The full reranking didn\'t complete. If partial results are tolerated the reranking got reverted and all documents preserved their original score and ranking.): Query limits: [TimeAllowedLimit:LIMIT EXCEEDED]'
  2>  response: {
  2>   "responseHeader":{
  2>     "partialResults":true,
  2>     "status":0,
  2>     "QTime":406,
  2>     "params":{
  2>       "fv":"true",
  2>       "q":"_query_:{!edismax qf='id' v='8^=10 9^=5 7^=3 6^=1'}",
  2>       "indent":"on",
  2>       "fl":"id,score",
  2>       "timeAllowed":"300",
  2>       "rows":"4",
  2>       "wt":"json",
  2>       "partialResults":"false",
  2>       "rq":"{!ltr model=slowModel reRankDocs=3}"
  2>     }
  2>   },
  2>   "response":{
  2>     "numFound":0,
  2>     "start":0,
  2>     "maxScore":0.0,
  2>     "numFoundExact":true,
  2>     "docs":[ ]
  2>   }
  2> }

[cpoerschke]

Unable to reproduce the TestLTRQParserPlugin.ltr_expensiveFeatureRescoringAndPartialResultsNotTolerated_shouldRaiseException issue locally. Nothing obvious from inspecting the CI output. Mailing list archive shows that occasionally this has happened before too: https://lists.apache.org/[email protected]:lte=1y:%22org.apache.solr.ltr.TestLTRQParserPlugin.ltr_expensiveFeatureRescoringAndPartialResultsNotTolerated_shouldRaiseException%22

[sigram]

I wonder whether we should add some information to the response header when multiThreaded=true is explicitly set in the params but then it's internally disabled due to the other conditions (limits, post filters, etc..) because it's not obvious anymore if MT was actually used as requested.

[dsmiley]

First started failing April 8th. SOLR-13350 landed May 6th.

I agree that we need clear diagnostics on when segments are actually searched in parallel or not.

[cpoerschke]

closing in favour of #2595 adding support for query limits;

separately #2596 to add documentation and responseHeader.multiThreadedUsed diagnostics;

[gus-asf]

First started failing April 8th. SOLR-13350 landed May 6th.

I agree that we need clear diagnostics on when segments are actually searched in parallel or not.

There actually are two log messages that indicate this but they were cryptic I updated them in #2595 here: https://github.com/apache/solr/pull/2595/files#diff-99978700f1c69d6a5f6c2190f89c98cfe20c441161db5a183ec002e15cb1be28R1924

[cpoerschke]

... they were cryptic I updated them in #2595 ...

I wonder if log level diagnostics might be enough, perhaps the diagnostic could be debug generally and info when the multiThreadedUsed contradicts multiThreaded in the request. A responseHeader element seems a bit messy in the code base -- see #2596 -- and also it seems impossible when the request was served from the cache.