RE: [Webmacro-devel] Bad interaction with TimeLoop and cache time s?

[Jeff C. <jef...@av...>]

Lane, thanks for the info. We'll work it into the next set of tests that we
do.

Can you point me at some documentation about how the
GenerationalCacheManager works or what it's key / differentiating features
are?

Regards,
Jeff

-----Original Message-----
From: Lane Sharman [mailto:la...@op...]
Sent: Tuesday, January 14, 2003 8:08 PM
To: Jeff Cowan
Cc: 'web...@li...'
Subject: Re: [Webmacro-devel] Bad interaction with TimeLoop and cache
times?


Hi Jeff,

One of the many beauties of WebMacro is pluggability.

About six months ago, we wrote a pluggable interface for the caching 
after some really hot debate about synchronization. I have been using 
exclusively an alternative cache manager which stems from my early days 
writing VM paging and kernel systems. It is a generational caching 
manager which can be configured as follows in webmacro.properties:

==========================
CachingProvider.template.CacheManager: 
org.webmacro.resource.GenerationalCacheManager
CachingProvider.url.CacheManager: 
org.webmacro.resource.GenerationalCacheManager
# Set reloading to true since we are in development mode by default.
GenerationalCacheManager.*.ReloadOnChange: true
# The refresh rate for incorporating changes to the mutable image
GenerationalCacheManager.*.RefreshRate: 1000
==========================

Would you mind copying the above into your WebMacro.properties file and 
seeing if you get some better performance or more control over your 
memory utilization and let me know. (Change the last parameter measured 
in millis to extend your mileage :).

Gracie,

Lane

Jeff Cowan wrote:

>Hello all. My name is Jeff and I work with Glenn Shute at Altavista. He
>suggested I drop a note here about something I've been looking at in
>WebMacro. We're running WM 1.0.1 still, so maybe this is already fixed or a
>warning has been put out already.
>
>I observed during some tests (memory profiling) that some templates were
>being parsed repeatedly when they should have already been cached. I
checked
>my cache timeout settings in WebMacro.defaults. They are:
>
>SMapCacheManager.*.ExpireTime: 600000
>SMapCacheManager.*.ReloadOnChange: false
>
>So, I would expect that once a template was acquired successfully, it
should
>be cached for 10 minutes. Looking more deeply, it appears that the TimeLoop
>class that is responsible for purging the template cache of stale templates
>has 600 periods of 1000ms each (again, 10 minutes).
>
>Setting LogLevel.resource to DEBUG, I observed that the template would be
>aquired, cached and subsequently deleted from the cache almost right away.
>It appears that when the period to run the task in is being computed in my
>particular case, it is possible that the period in question is the one that
>is about to run right at that moment. This would only happen in the case
>where the periods corresponded to the ExpireTime, due to modulo arithmetic.
>Eventually, if I continue reloading the page, all the templates get cached
>successfully. This kind of error is a subtle timing issue, so the results
>are different every time. 
>
>In our case, we can change the ExpireTime to 599000 and that seems to fix
>it. I understand that this isn't really a bug. Just thought you might be
>interested.
>
>Regards,
>Jeff Cowan
>Altavista Company
>email: jeff <d o t> cowan <a t> av <d o t> com
>
>
>
>
>
>-------------------------------------------------------
>This SF.NET email is sponsored by: Take your first step towards giving 
>your online business a competitive advantage. Test-drive a Thawte SSL 
>certificate - our easy online guide will show you how. Click here to get 
>started: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0027en
>_______________________________________________
>Webmacro-devel mailing list
>Web...@li...
>https://lists.sourceforge.net/lists/listinfo/webmacro-devel
>
>  
>

-- 
Lane Sharman
http://opendoors.com Conga, GoodTimes and Application Hosting Services
http://opendoors.com/lane.pdf BIO