Date: Tue, 11 Mar 1997 14:55:41 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: joerg_wunsch@uriah.heep.sax.de Cc: freebsd-hackers@freebsd.org Subject: Re: Hard Link Count too small! Message-ID: <199703112155.OAA26178@phaeton.artisoft.com> In-Reply-To: <19970311202555.KX24648@uriah.heep.sax.de> from "J Wunsch" at Mar 11, 97 08:25:55 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > Yes. And this is what must be prevented. POSIX compliance must be > > maintained by stable storage, ... > > So any system without RAID would be in danger of losing POSIX > compliance under some circumstances. Only if you refuse to provide atomicity for non-idempotent operations against the FS. Like if you mount the thing -async. Otherwise, it is deterministically recoverable to the correct state, as mandated by POSIX, since it is done one-behind. The VIVA FS papers provide a wonderful study of the application of idempotence requirements in the context of the POSIX standards. There is a potential disconnect in the soft updates case for any system (like FreeBSD) which unifies the VM and buffer cache. For those systems, soft updates will degrade to DOW performance in order to make the guarantees, unless soft updates are implemented at the FS/BIO (in the FreeBSD case, FS/VM) interaction layer instead of in the FS proper. This is pretty obvious from the Ganger/Patt paper, for what it's worth. Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199703112155.OAA26178>