You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-issues@jackrabbit.apache.org by "Nuno Santos (Jira)" <ji...@apache.org> on 2024/09/27 09:20:00 UTC
[jira] [Resolved] (OAK-11131) indexing-job: AOT Blob downloader may download blobs that are not needed for the indexes
[ https://issues.apache.org/jira/browse/OAK-11131?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Nuno Santos resolved OAK-11131.
-------------------------------
Fix Version/s: 1.72.0
Resolution: Done
> indexing-job: AOT Blob downloader may download blobs that are not needed for the indexes
> ----------------------------------------------------------------------------------------
>
> Key: OAK-11131
> URL: https://issues.apache.org/jira/browse/OAK-11131
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: indexing
> Reporter: Nuno Santos
> Priority: Minor
> Fix For: 1.72.0
>
>
> This is the current logic to decide to download a blob:
> - AOT Blob downloader must be enabled for one of the indexes being indexed
> - The node name must end with the suffix configured in AOT blob downloader
> - The node must be indexed by one of the indexers
> This will result in downloading too many blobs. Consider that we are reindexing fooBar-1, which includes path {{/content/dam}} and some other index fooBaz-1 that covers the full repository, not only {{/content/dam}}. AOT blob download is enabled for fooBar-1. Then any node appearing outside {{/content/dam}} that matches the suffix, will also pass all the tests and its blob will be downloaded, even though fooBar does not need the blob.
> The fix is to check if the node being considered for blob download is indexed by the indexes for which AOT download is enabled, instead of checking against all the indexes.
> This issue should not cause any failures. It makes the indexing job download more than it should and may slow down the indexing phase, but likely not slower than it would be without AOT Blob download. So It is not a blocker to use this feature, but should be fixed for better performance and avoid useless work.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)