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
_______________________________________________
squid-users mailing list
squid-users@xxxxxxxxxxxxxxxxxxxxx
http://lists.squid-cache.org/listinfo/squid-users
