Re: cache_dir size weirdness

On Wed, 29 Sep 1999, Chris Dillon wrote:

>
> I'm running 2.2-STABLE4 (previously STABLE3) on FreeBSD 3.3-STABLE and
> I just noticed something odd about my cache. From the beginning, I've
> had my cache directories set up like so:
>
> cache_dir /cache.0 6000 16 256
> cache_dir /cache.1 6000 16 256
> cache_dir /cache.2 6000 16 256
>
> But I've noticed that each cache_dir never seems to grow past about
> 1.2GB of usage.
>
> Filesystem 1K-blocks Used Avail Capacity Mounted on
> /dev/da0s1g 6625221 1251550 5307419 19% /cache.0
> /dev/da1s1g 6625221 1252212 5306757 19% /cache.1
> /dev/da2s1g 6625221 1252062 5306907 19% /cache.2
>
> All this time, the following line in the logs has been escaping my
> eye:
> Sep 29 10:01:56 cheetah squid[40163]: store_swap_size = 3697752k
>
> Does this log entry specify how much data is currently in the cache,
> or the maximum cache size?
>
> I don't think anything else in my config could affect the cache
> directories and how much they can hold, except possibly a larger
> maximum_object_size of 8192K.
>
> Any ideas?

There is another factor. Check the config parameter called
'reference_age'. Objects older than this value are removed.

You might check that value, then look for LRU in cachemgr "info":

squid/src 221> ./client mgr:info|grep LRU
        Storage LRU Expiration Age: 3.78 days

If the two values are the same, then Squid is removing objects even
though your cache capacity hasn't been reached.

Duane W.
Received on Wed Sep 29 1999 - 11:43:30 MDT