Date: Wed, 14 Jan 2004 10:37:37 -0800 From: "Kevin Oberman" <oberman@es.net> To: John Baldwin <jhb@FreeBSD.org> Cc: current@freebsd.org Subject: Re: Serious dd breakage in current Message-ID: <20040114183737.6C0105D08@ptavv.es.net> In-Reply-To: Message from John Baldwin <jhb@FreeBSD.org> of "Tue, 13 Jan 2004 15:04:14 EST." <200401131504.15194.jhb@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> From: John Baldwin <jhb@FreeBSD.org> > Date: Tue, 13 Jan 2004 15:04:14 -0500 > > On Monday 12 January 2004 07:19 pm, Dan Nelson wrote: > > In the last episode (Jan 12), Kevin Oberman said: > > > Today I bit the bullet and re-sized some partitions on my laptop's > > > disk. One think I planned to do was copy the unchanged partitions from > > > my backup disk to the primary with dd(1). This was a BAD idea and I > > > suspect GEOM changes are at the root of it. > > > > > > I used fdisk to create new slices and then bsdlabel to make new > > > partitions in ad0s2. Everything seemed to be fine. > > > > > > Then I ran dd to copy the root partition over: dd bs=32k if=/dev/ad2s3a > > > of=/dev/ad0s2a For some reason it labeled the disk with the first > > > partition starting at almost the end of the physical partition, over 30 > > > million blocks into the slice. bsdlabel generated a stream of errors > > > including that every partition extended past the physical partition. > > > > You can't dd disklabel partitions, as they use absolute offsets from > > the start of the disk, not the start of the slice. The disklabel > > command compensates for this so you see relative offsets. You can work > > around the problem by dumping your partition info to a text file and > > updating the new disk with disklabel after the dd command. > > dump / restore is a better method though anyways. It doesn't bother with > copying unused blocks for one thing. It will also allow things like dirhash > to more efficiently lay out your files on the new file system. > > -- > John Baldwin <jhb@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/ > "Power Users Use the Power to Serve" = http://www.FreeBSD.org > Thanks to both John and Dan. I clearly did not correctly understand how dd operated. Guess it's time to read the sources a bit. I understand why dump|restore is a better choice than dd in may ways, but, if a partition is large and full, dd is MUCH faster. That's why I use it to backup my system disk. I can copy 40 GB in about 40 minutes on my laptop and dumping takes just a bit longer. Non the less, now that I see the problem with dd on a partition, I have used dump and my system is now properly re-partitioned. It's nice to see both /var and /usr under 98%! Thanks again for taking the time to explain this. -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: oberman@es.net Phone: +1 510 486-8634
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040114183737.6C0105D08>