[squid-users] What would be the maximum ufs\aufs cache_dir objects?

[Amos Jeffries]

On 15/07/17 00:37, Eliezer Croitoru wrote:
> What would be the maximum ufs\aufs cache_dir objects? > Let say I have unlimited disk space and inodes and RAM, what would be the
> maximum objects I can store on a single ufs\aufs cache_dir?


One UFS cache_dir can hold a maximum of (2^27)-1 safely.

Technically it does not need the -1, but the old C code uses a mess of 
signed and unsigned types to store the has ID value. Some (not all) 
people hit assertions when the cache reaches that boundary.


> It's very easy to test but first I want to understand what might be the
> limit?

The index hash entries are stored as a 32-bit bitmask (sfileno) - with 5 
bits for cache_dir ID and 27 bits for hash of the file details.


> I am asking since the structure is top level dirs and sub level dirs, so if
> I want to get the maximum object capacity (assuming each one would use
> 0.5kb)?

The L1/L2 separation is to cope with old filesystems that had limited 
number of files in a directory.

Apparently, that limitation is no longer relevant with the current 
generation of filesystems.


Amos