Comments on capricious diatribes: Hibernate Query Cache: A Dirty Little Secret

Great article. This gave me the real taste and com...

2011-07-29T04:19:59.670-05:00

Great article. This gave me the real taste and complexity in understanding caching in ORM.
Thank you for posting this.

I hope you will keep updating your content constan...

2011-04-10T13:18:44.732-05:00

I hope you will keep updating your content constantly as you have one dedicated reader here.

cheap clomid

Great work keep it coming, best blog on earth
2011-04-08T16:04:22.485-05:00
Great work keep it coming, best blog on earth

aromatherapy world

The bug 3383 seems to have been fixed in Hibernate...

2009-06-16T15:20:36.659-05:00

The bug 3383 seems to have been fixed in Hibernate 3.2.7 which will be released in weeks, if not days.

Interestingly there is another bug in hibernate HH...

2009-06-05T14:26:16.834-05:00

Interestingly there is another bug in hibernate HHH-3028. This is causing huge memory leak if the query cache is enabled. Even if we evict the objects/or clear the objects after isert/update still hibernate holds the objects in memory when query cache is enabled.

and once again, i got caught not looking at those ...

2009-05-12T11:47:00.000-05:00

and once again, i got caught not looking at those particular sources myself. :\

Hi Steve, That's good to know. I had the impressi...

2009-05-12T06:43:00.000-05:00

Hi Steve,

That's good to know. I had the impression that I'd have to crawl through many QueryKey references scattered through the codebase. Does this mean we should expect an official fix soon?

BTW, what the Hibernate release status? It's been a while.

QueryKey is built in a single place org.hibernate....

2009-05-11T23:15:00.000-05:00

QueryKey is built in a single place org.hibernate.loader.Loader#listUsingQueryCache, which does have access to the session...

d'oh! i didn't actually look where the QueryKey c...

2009-05-11T21:33:00.000-05:00

d'oh! i didn't actually look where the QueryKey constructor was used. i was looking at QueryCache.put() which takes both the query key AND the SessionImplementor. which, would make it easy in my 'decorator' implementation to get the id from an entity. but it doesn't help in the generic "fix the QueryKey" case. rats.

Hi Olag, I did actually consider Type.disassemble...

2009-05-11T19:06:00.000-05:00

Hi Olag,

I did actually consider Type.disassemble() however it requires access to a SessionImplementor which is not available within the scope of QueryKey. It could be passed as a parameter but this would require modifying many other source classes. The proposed solution is self-contained (assuming it works as intended).

@Manuel -- I looked at your sample java but didn't...

2009-05-11T17:09:00.000-05:00

@Manuel -- I looked at your sample java but didn't get a chance to try it. I think we're on the right track... but there's been some traffic on a newer post with some comments from a hibernate dev. He suggested we look at the Type.disassemble() method since we have Type available for both named and positional parameters. For Entity types, it should returned the serializable identifier. And for other types, it should "do the right thing."

Well it got mangled again - the file is QueryKey.j...

2009-05-06T21:25:00.000-05:00

Well it got mangled again - the file is QueryKey.java (java extension)

Here's the link again - http://www.residencialagua...

2009-05-06T21:24:00.000-05:00

Here's the link again - http://www.residencialaguardia.com/temp/QueryKey.java for some reason it got mangled in the previous post.

Check out this possible solution http://www.reside...

2009-05-06T21:23:00.000-05:00

Check out this possible solution http://www.residencialaguardia.com/temp/QueryKey.java

It's a patched QueryKey which attempts to detect whether an object is an entity and replaces its value with its key. Should work for both positional and named parameters as well.

It's not tested at all - I just did this as a late-night proof-of-concept exercise. It uses Apache Commons BeanUtils for reflection but it could easily be done without it.

Let me know what you think. If you find it works well we can give it a better shape and try to contribute back to Hibernate.

Agreed, on fixing internally to hibernate. I don'...

2009-05-06T09:46:00.000-05:00

Agreed, on fixing internally to hibernate. I don't claim to be an expert -- so maybe there is cache/JVM clustering issues I am not aware of.

A newer post details some more duplication waste I came across --

Query Cache Strikes Back

Hi there, I posted the original JIRA bug report. S...

2009-05-05T06:53:00.000-05:00

Hi there, I posted the original JIRA bug report. Soon after posting I also realized that we could rewrite criteria queries as well and this is in fact what we did for all our cacheable queries. However the real solution is fixing the Hibernate core - new developers will most likely ignore this subtle problem and this would go unnoticed until the apps go into production and start eating memory up.

Nice blog entry. I would approach the solution by storing only primitives in the QueryCacheKey, both for positional and named parameters. That's fool-proof if we can pull it off elegantly.

Thanks for the kind words. Although I wouldn't go...

2009-04-14T22:51:00.000-05:00

Thanks for the kind words. Although I wouldn't go back to implementing my own ORM, I'll admit hibernate continues to "surprise" me quite often...

Wonderful article. I would say you have done a ver...

2009-03-18T02:50:00.000-05:00

Wonderful article. I would say you have done a very good job in explaining how query caching works and its limitations.