[hibernate-issues] [Hibernate-JIRA] Commented: (HHH-4550) query cache: Attention with configuring Update Timestamps Cache region: expired entries cause wrong result sets!

[Guenther Demetz (JIRA)]

    [ http://opensource.atlassian.com/projects/hibernate/browse/HHH-4550?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=34496#action_34496 ] 

Guenther Demetz commented on HHH-4550:
--------------------------------------

Finally in the JavaDoc of class UpdateTimestampsCache I have found more less the advice I searched for:

/**
 ... 
 * It is important that the cache timeout of the underlying cache implementation be set
 * to a higher value than the timeouts of any of the query caches. In fact, we
 * recommend that the the underlying cache not be configured for expiry at all.
 * Note, in particular, that an LRU cache expiry policy is never appropriate.
 ...
 */
public class UpdateTimestampsCache ...


Personally I think this hint should appear in every article speaking about configuring hibernate query cache.

> query cache: Attention with configuring Update Timestamps Cache region: expired entries cause wrong result sets!
> ----------------------------------------------------------------------------------------------------------------
>
>                 Key: HHH-4550
>                 URL: http://opensource.atlassian.com/projects/hibernate/browse/HHH-4550
>             Project: Hibernate Core
>          Issue Type: Improvement
>          Components: caching (L2)
>    Affects Versions: 3.3.2
>         Environment: 3.3.2 GA
>            Reporter: Guenther Demetz
>            Priority: Trivial
>         Attachments: TestCaseQuryCache.jar
>
>
> When using hibernate's query cache (hibernate.cache.use_query_cache),
> then it's of fundamental importance to configure the Update Timestamps cache region in
> a way that his entries survive longer than the cached result sets in the Query cache region.
> Otherwise queries could return wrong (obsolete) result sets (See attached testcase).
> This is because the current UpdateTimestampsCache#isUpToDate implementation considers a result-set also as up-to-date,
> if there are no update-timestamps cached for the interested spaces (tables) at all.
> It if therefore important to notice that:
> 1- The Update-Timestamps-Cache-Region elements max-size should be configured higher than the number of tables you have 
>    on the db-schema, in manner that he can remember/hold the last update-timestamp for each table.
> 2- If you use any other eviction policy in the Update-Timestamps-Cache-Region, then you must assure 
>    that cached result sets in the Query-cache-region are evicted before relevant update-timestamps are going to be evicted.
> This fact should be clearly documented somewhere.
> Thanks
> G.D.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira