Hello!

On Thu, Jul 18, 2013 at 08:54:48PM -0400, feanorknd wrote:

> Thanks thanks so much!!! :D
>
> I even saw that ticket before posting, but I figured out it was not the
> problem just because I do use XFS for my nginx_caches at 4 servers without
> this problem, and also I did test changing the path to point to a XFS
> partition on SATA drive, and the problem dissapeared.
>
> If you have a look at the ticket, that user also use a SSD drive and have
> the alloc problem.... it seems like if only at SSD drives, the size
> notification is incorrect because of the allocsize, but not on SATA drives,
> although the allocsize is default for both of them, at least in my case!!!!
>
> Have a look at my first post and see at SATA drive, with XFS defaults, the
> cache gets fullfilled exactly and if I configure 2500Gb, the "du -hs" when
> core starts force expiring objects, is just that 2500Gb..... the measures al
> exact and real.
>
> Not the same for SSD.. do you think people is something to review by
> developers somehow?

Depending on media used XFS might apply different defaults, and/or
observed behaviour might be different due to timing reasons.
Quick search suggests XFS currently uses dynamic allocsize by
default, see here:

http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=055388a3188f56676c21e92962fc366ac8b5cb72

Try forcing allocsize to something like 4k to see if it helps.

--
Maxim Dounin
http://nginx.org/en/donation.html

_______________________________________________
nginx mailing list
nginx@nginx.org
http://mailman.nginx.org/mailman/listinfo/nginx