[
https://issues.apache.org/jira/browse/JENA-118?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13109539#comment-13109539
]
Damian Steer commented on JENA-118:
-----------------------------------
Unlocking occurs at the end of the transaction in postgresql. Other databases
(mysql, db2, oracle) have something similar, but (surprise!) details vary.
MySQL appears to need an explicit unlock.
I should clarify the 'heavy' bit: it holds the lock longer than it strictly
needs to (after node table modification is complete), and I think it prevents
'safe' concurrent reads, i.e. sparql queries. However it enables concurrent
(ish :-) writes, so I'm inclined to enable it. Original reporter doesn't seem
to have tried it.
> SDB can't concurrently write (duplicate key error)
> --------------------------------------------------
>
> Key: JENA-118
> URL: https://issues.apache.org/jira/browse/JENA-118
> Project: Jena
> Issue Type: Improvement
> Components: SDB
> Reporter: Damian Steer
> Assignee: Damian Steer
> Attachments: lock-pg.patch
>
> Original Estimate: 96h
> Remaining Estimate: 96h
>
> Multiple concurrent writes in SDB can break because duplicates can appear
> despite the join-based duplicate suppression. Typically the users sees
> something like:
> org.postgresql.util.PSQLException: ERROR: duplicate key value violates unique
> constraint "nodes_pkey"
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
