> Hi! > > We have a reverse squid proxy running fine. But we need to put the > cache_replacement_policy in mode GDFS. For this, we have this in > config file: > > cache_replacement_policy heap GDSF > memory_replacement_policy heap GDSF > high_memory_warning 2048 MB > > With independence about the convenience of using GDFS with > cache_replacement_policy and memory_replacement_policy, when we run > the command: > > $ squidclient -p 80 mgr:storedir > > HTTP/1.0 200 OK > Server: squid > Mime-Version: 1.0 > Date: Mon, 02 Mar 2009 16:09:50 GMT > Content-Type: text/plain > Expires: Mon, 02 Mar 2009 16:09:50 GMT > Last-Modified: Mon, 02 Mar 2009 16:09:50 GMT > X-Cache: MISS from vlex.eu > X-Cache-Lookup: MISS from vlex.eu:80 > Via: 1.0 vlex.eu (squid) > Connection: close > > Store Directory Statistics: > Store Entries : 6504719 > Maximum Swap Size : 41943040 KB > Current Store Swap Size: 39530824 KB > Current Capacity : 94% used, 6% free > > Store Directory #0 (aufs): /var/cache/squid > FS Block Size 4096 Bytes > First level subdirectories: 16 > Second level subdirectories: 64 > Maximum Size: 41943040 KB > Current Size: 39530824 KB > Percent Used: 94.25% > Filemap bits in use: 6503017 of 8388608 (78%) > Filesystem Space in use: 39971720/138299856 KB (29%) > Filesystem Inodes in use: 6584740/8781824 (75%) > Flags: SELECTED > Removal policy: lru > LRU reference age: 1.88 days > > As you can see, the Removal policy is marked as "lru", not GDFS. Why? > We are a bit confusing about this. Many things in squid.conf are order-specific. The cache_replacement_policy setting can be used multiple times, and only affects the cache_dire defined below it. Amos
