> matthew green <mrg%eterna.com.au@localhost> wrote: > > > this seems like the wrong answer. if rlimits aren't enough, then the > > *user* should be increasing them, not the system. > > The problem is that we have no way to pass that as a mount option, and > no way to evaluate how much memory will b e required. a cache should have _no_ requirement. just a benefit. > > additionally, why does it have to have so much memory? surely most of > > it is a cache? can't it manage the size? > > Indeed it is cache. You cannot expect filesystems to let you control > that. Speaking of FUSE it seems they will not let you do it since Linux > data size is unlimited by default. none of this is any good reason to overrride the users rlimit. the perfuse code should maintain a cache that keeps under the users data rlimit. this change should be reverted. .mrg.