Re: Question on TemplateCache synchronization

I just finished testing trunk with our performance test suite and the
gains are significant (though there was also one other fix in the test
with another overly synchronized block of code in an often called
class that also would have had a significant impact). I'm wondering
what the timetable for tagging this as 2.3.13 ?


Regards,

Bruce Ritchie

On Wed, Apr 9, 2008 at 9:05 PM, Daniel Dekany <[email protected]> wrote:
> Is this just a coincidence that you ask it now? Because, the reworked
>  caching system was committed only like days ago. It's much more
>  multi-core friendly now. Grab the HEAD of the 2.3 branch from
>  https://freemarker.svn.sourceforge.net/svnroot/freemarker/branches/2.3/freemarker
>  and give it a try. (I wanted to say to grab the latest nightly build
>  jar, but it seems that service is down at the moment.)
>
>
>
>  Thursday, April 10, 2008, 12:19:06 AM, Bruce Ritchie wrote:
>
>  > I'm currently analyzing a thread dump from a production environment
>  > where I have 100 threads in waiting on the CacheStorage lock in
>  > TemplateCache:
>  >
>  > at freemarker.cache.TemplateCache.getTemplate(TemplateCache.java:246)
>  > - waiting to lock <0x27445a20> (a freemarker.cache.MruCacheStorage)
>  > at freemarker.cache.TemplateCache.getTemplate(TemplateCache.java:229)
>  > at
>  > freemarker.template.Configuration.getTemplate(Configuration.java:487)
>  > at
>  > freemarker.template.Configuration.getTemplate(Configuration.java:452)
>  >
>  > and one thread actually holding the lock:
>  >
>  > at java.lang.Throwable.fillInStackTrace(Native Method)
>  > at java.lang.Throwable.<init>(Throwable.java:196)
>  > at java.lang.Exception.<init>(Exception.java:41)
>  > at javax.naming.NamingException.<init>(NamingException.java:109)
>  > at
>  > org.apache.naming.resources.FileDirContext.lookup(FileDirContext.java:208)
>  > at
>  > org.apache.naming.resources.ProxyDirContext.lookup(ProxyDirContext.java:294)
>  > at
>  > org.apache.catalina.loader.WebappClassLoader.findResourceInternal(WebappClassLoader.java:1891)
>  > at
>  > org.apache.catalina.loader.WebappClassLoader.findResource(WebappClassLoader.java:934)
>  > at
>  > org.apache.catalina.loader.WebappClassLoader.getResource(WebappClassLoader.java:1069)
>  > at
>  > com.jivesoftware.community.util.ChainingClassLoader.getResource(ChainingClassLoader.java:91)
>  > at
>  > com.jivesoftware.community.web.struts.JiveClassTemplateLoader.getURL(JiveClassTemplateLoader.java:27)
>  > at
>  > freemarker.cache.URLTemplateLoader.findTemplateSource(URLTemplateLoader.java:75)
>  > at
>  > freemarker.cache.MultiTemplateLoader.findTemplateSource(MultiTemplateLoader.java:111)
>  > at
>  > freemarker.cache.TemplateCache.acquireTemplateSource(TemplateCache.java:520)
>  > at
>  > freemarker.cache.TemplateCache.findTemplateSource(TemplateCache.java:510)
>  > at freemarker.cache.TemplateCache.getTemplate(TemplateCache.java:315)
>  > - locked <0x27445a20> (a freemarker.cache.MruCacheStorage)
>  > at freemarker.cache.TemplateCache.getTemplate(TemplateCache.java:229)
>  > at
>  > freemarker.template.Configuration.getTemplate(Configuration.java:487)
>  > at
>  > freemarker.template.Configuration.getTemplate(Configuration.java:452)
>  >
>  >
>  > Is there really a good reason for synchronizing on the CacheStorage?
>  >>From my reading of the code there might be a small race between
>  > storage.get()l and storage.remove() but I'm not sure it's really worth
>  > synchronizing for. Am I wrong in this statement? If so, can someone
>  > suggest an alternate approach that doesn't require synchronization? In
>  > production we set the templateUpdateDelay to Integer.MAX_VALUE so
>  > basically we should never have to worry about the template changing or
>  > being removed after it's loaded.
>  >
>  >
>  > Regards,
>  >
>  > Bruce Ritchie
>  >
>  > -------------------------------------------------------------------------
>  > This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
>  > Don't miss this year's exciting event. There's still time to save $100.
>  > Use priority code J8TL2D2.
>  > http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
>  > _______________________________________________
>  > FreeMarker-user mailing list
>  > [email protected]
>  > https://lists.sourceforge.net/lists/listinfo/freemarker-user
>
>  --
>  Best regards,
>   Daniel Dekany
>
>

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone