[
https://issues.apache.org/jira/browse/OAK-2813?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14516565#comment-14516565
]
Marcel Reutegger commented on OAK-2813:
---------------------------------------
{noformat}
15:50:07.923 [oak-scheduled-executor-8] DEBUG o.a.j.o.p.index.AsyncIndexUpdate
- [async] AsyncIndex update run completed in 23.54 s. Indexed 16270 nodes
15:51:15.890 [oak-scheduled-executor-1] DEBUG o.a.j.o.p.index.AsyncIndexUpdate
- [async] AsyncIndex update run completed in 1.049 min. Indexed 38088 nodes
15:53:25.757 [oak-scheduled-executor-13] DEBUG o.a.j.o.p.index.AsyncIndexUpdate
- [async] AsyncIndex update run completed in 2.081 min. Indexed 108368 nodes
15:58:05.321 [oak-scheduled-executor-5] DEBUG o.a.j.o.p.index.AsyncIndexUpdate
- [async] AsyncIndex update run completed in 4.493 min. Indexed 192446 nodes
{noformat}
I think Oak-Tar is able to add more nodes than the AsyncIndexUpdate is able to
process. E.g. the index update processed 108368 nodes in 2.081 min, but during
that time, the import added another 192446, as seen in the next round of the
index update. It looks like the import adds roughly twice as many nodes as the
index updater is able to process, which means the importer probably added
another 400k nodes during the 4.493 min index update.
> Create a benchmark for measuring the lag of async index
> -------------------------------------------------------
>
> Key: OAK-2813
> URL: https://issues.apache.org/jira/browse/OAK-2813
> Project: Jackrabbit Oak
> Issue Type: Sub-task
> Components: core
> Reporter: Davide Giannella
> Assignee: Davide Giannella
> Fix For: 1.3.0
>
> Attachments: OAK-2813-r1.diff, benchmark-run.log,
> logback-benchmark.xml
>
>
> As with OAK-2749 (and others) we're taking routes for optimising the speed of
> async indexing, we need a benchmark that will measure how long it takes for a
> lucene property index to have the content indexed.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
