On 17/05/11 21:34, Tom Tux wrote:
Actual, I have 69% free storedir-capacity (62GB) The filemap shows: Filemap bits in use: 1805414 of 2097152 (86%) Is there a way to increase the file filemap-max-value to go around the purging-steps? Is it depending on the numbers of cache_dir-subdirectories?
This is about absolute cache file count not disk bytes. There is no way to increase that filemap. UFS in Squid has an absolute 32-bit number used to index each cache_dir, with a few bits lost to overheads.
You can tune the cache_dir avg object size via min-size/max-size options and thus alter the avg bytes per filemap entry used to better utilize the disk space.
If you have squid-2.7 you can also add a COSS directory in any spare space next to a UFS/AUFS/DiskD drive. That allows a whole separate fileno index for small files stored in the COSS, large files in the *UFS cache_dir.
With ~70% of the drive used for ~85% of the cache_dir capacity you are doing well. That will give you a bit of leeway for the swap.state journals and some 10% bytes overflow the disk FS often needs to work efficiently.
Amos -- Please be using Current Stable Squid 2.7.STABLE9 or 3.1.12 Beta testers wanted for 3.2.0.7 and 3.1.12.1