Date: Tue, 26 Feb 2008 19:45:33 +0100 From: "Martin Laabs" <martin.laabs@mailbox.tu-dresden.de> To: "freebsd-hackers@freebsd.org" <freebsd-hackers@freebsd.org> Subject: Re: emulate an end-of-media Message-ID: <op.t65ad7g2724k7f@martin> In-Reply-To: <20080226075102.GE83599@server.vk2pj.dyndns.org> References: <op.t63j2veq724k7f@martin> <20080225154455.4822e72a@bhuda.mired.org> <20080226075102.GE83599@server.vk2pj.dyndns.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi, On Tue, 26 Feb 2008 08:51:02 +0100, Peter Jeremy <peterjeremy@optushome.com.au> wrote: > [...] But it also has some gotchas. The > biggest one is that dump needs to be able to write past EOM (so it can > record an end-of-volume block). Just even I tried the following: -create a 50MB filesystem (with md) -put 29MB data on it -create 3 10MB md-devices -dump the 50MB filesystem at the three 10MB devices -restore the file into another, fresh 50MB filesystem Restore reported no error, all files are there and the number of used inodes is the same. Also the md5 checksum above all files does not differ. Since I'm almost sure that the md-devices does not allow to write behind the EOM signal this test would be an equvilalent to the SIGPIPE to SIGEOM thing. (Regardless how I implement this at the end.) So it doesn't seem that dump can handle such an abrupt EOM. This evening I'll try to change dump to interprete SIGPIPE like SIGEOM. Best regards, Martin L.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?op.t65ad7g2724k7f>