I updated the docs, thanks for the patch.

On Tue, May 29, 2007 at 09:32:06AM +0200, BUSTARRET, Jean-francois wrote:
> 
> As far as I understand the code, refcount is the number of concurrent 
> requests working on an item.
> Something from the tail with refcount > 0 is being accessed and will soon be 
> moved to the head of the LRU.
> 
> The doc is wrong. Here is what it should say :
> 
> diff -rup memcached-1.2.2/doc/protocol.txt 
> memcached-1.2.2-doc/doc/protocol.txt
> +++ memcached-1.2.2-doc/doc/protocol.txt        2007-05-29 09:19:12.000000000 
> +0200
> @@ -352,8 +352,8 @@ get_hits          64u      Number of key
>                             found present
>  get_misses        64u      Number of items that have been requested
>                             and not found
> -evictions         64u      Number of items removed from cache because
> -                           they passed their expiration time
> +evictions         64u      Number of valid items removed from cache
> +                           to free memory for new items
>  bytes_read        64u      Total number of bytes read by this server
>                             from network
>  bytes_written     64u      Total number of bytes sent by this server to
>  
> Can someone with enough karma apply the patch ?
> 
> JFB
> 
> De : Ben Hartshorne [mailto:[EMAIL PROTECTED] 
> Envoyé : mardi 29 mai 2007 01:19
> À : BUSTARRET, Jean-francois
> Cc : [email protected]
> Objet : Re: evictions stat in 1.2.2
> 
> On Mon, May 28, 2007 at 05:32:15PM +0200, BUSTARRET, Jean-francois wrote:
> > 
> > The eviction stat only counts "cache overflows" (valid entries bumped 
> > from the cache by inserts) and not expirations.
> > 
> > What would be the point of counting expirations ?
> 
> I agree that I am more interested in cache overflows.  However, the docs in 
> 'protocol.txt' included with the tarball describes:
> 
>     355 evictions         64u      Number of items removed from cache because
>     356                            they passed their expiration time
> 
> This seems incorrect, both by my reading of the code and your statement 
> above.  Unfortunately, I can't quite describe exactly what evictions are 
> because I don't understand the conditions under which an item's refcount can 
> be nonzero (and therefore not bumped to make way for new items).
> 
> -ben
> 

-- 
Paul Lindner        ||||| | | | |  |  |  |   |   |
[EMAIL PROTECTED]

Attachment: pgpicqhWQV2zE.pgp
Description: PGP signature

Reply via email to