Re: maximum_object_size

On Tuesday 18 December 2001 16.34, Jon Kay wrote:

> Yes, but this doesn't solve that problem, it just muddles it. If somebody
> has a 2MB cache, this doesn't help. If somebody has a 2GB cache (pretty
> darned likely, actually), this doesn't help.

If someone has an cache in the range 50MB to 3GB it solves the problem.

If someone has an cache smaller than 50MB the problem will be very much
evident if there is any load, unless the maximum_object_size is lowered.

If someone has a cache larger than 3GB then the problem is pretty much hidden
even if the directive is removed as Squid cannot at all manage objects larger
than 2GB on 32 bit processors.

> Yes, it is. But like all 'max size' parameters, it has the problem
> of needing constant adjustment upward. 4MB was an appropriate amount
> four years ago. 400 MB would be a more appropriate number now, IMHO.
> Can we at least default it to not actively limiting anything?

If you feel happy with Squid swaping out 2GB objects to disk irregarless of
how large their cache actually is and then crashing because the disk got full
then so be it, but I don't see that as an option.

I can agree that the directive is perhaps muddling the real problem of how to
manage large objects without running out of space, but until the real problem
is fixed removing the limit will only make the situation worse.

Regards
Henrik
Received on Tue Dec 18 2001 - 08:45:20 MST