[
https://issues.apache.org/jira/browse/JENA-1987?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17225569#comment-17225569
]
Andy Seaborne commented on JENA-1987:
-------------------------------------
Yes - live compaction is possible.
One choice was fast-compaction, for example doing it like the TDB2 bulk loader
which throws a lot fo machinery out of the way to manipulate the datastructures
directly but as a consequence need exclusive access for an extended period of
time.
The live-compaction choice is a low-level but within the transaction manager
system. The bulk of the work is "block writers, a copy (read) to new database
(write)" and exclusive access only to switch from source to destination
database.
> HTTP API for TDB compact AND/OR periodical compact
> --------------------------------------------------
>
> Key: JENA-1987
> URL: https://issues.apache.org/jira/browse/JENA-1987
> Project: Apache Jena
> Issue Type: New Feature
> Components: Fuseki
> Reporter: Martin Kopeček
> Priority: Major
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> To be able to use Fuseki server as a standalone service, we need to run the
> compacting of the database from a different node (after updating the data in
> the database) using HTTP API.
> Other solution would be to implement runnign tdb compact on a configurable
> interval (e.g. 5 minutes).
> Our tdb database is updated every five minutes (so grows rapidly) and manual
> compacting is not an option.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
