On fre, 2008-07-04 at 04:35 -0700, John Doe wrote:
> Ahhh... when will it be written in the cache_dir and freed from RAM?
It's written out immediately, but not freed from ram as there is a
client waiting to get that data sent to it "immediately":
> The problem is that the squids will be used to cache videos.
> 100 clients asking for 100 different (not cached yet) xxMB videos will indeed exhaust the RAM...
I suspected so...
> Is there a way to avoid this scenario?
Modifying the code to behave as you like.
Today the original client is a bit special and can not get data from
disk, only memory. But this is not strictly required I guess.. (the
second+ client can get data from both as needed)
> More generally, when a client asks for a cached object; is the whole object loaded in RAM or is it loaded "block" by "block"?
"block by block".
Regards
Henrik
This archive was generated by hypermail 2.2.0 : Fri Jul 04 2008 - 12:00:02 MDT