[
https://issues.apache.org/jira/browse/LUCENE-8392?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16536789#comment-16536789
]
Adrien Grand commented on LUCENE-8392:
--------------------------------------
Here is a patch. It computes {{mergeFactor=min(segsPerTier, maxMergeAtOnce)}}
and uses it instead of maxMergeAtOnce.
> TieredMergePolicy has broken assumptions when maxMergeAtOnce is greater than
> segmentsPerTier
> --------------------------------------------------------------------------------------------
>
> Key: LUCENE-8392
> URL: https://issues.apache.org/jira/browse/LUCENE-8392
> Project: Lucene - Core
> Issue Type: Test
> Reporter: Adrien Grand
> Priority: Minor
> Attachments: LUCENE-8392.patch
>
>
> While working on LUCENE-8391 I had test failures when {{maxMergeAtOnce}} is
> larger than {{segmentsPerTier}}. For instance when all segments are on the
> same tier, the maximum number of segments that is allowed in the index is
> {{segmentsPerTier}} but because the tiered policy wants to find
> {{maxMergeAtOnce}} segments to merge, no segments will get merged if there
> are less than {{maxMergeAtOnce}} segments.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
