Date: Sun, 18 Sep 2005 11:16:18 +0700 From: Victor Sudakov <sudakov@sibptus.tomsk.ru> To: Danny Howard <dannyman@toldme.com> Cc: freebsd-geom@freebsd.org Subject: Re: Create a mirror on disk with valid data Message-ID: <20050918041618.GA59139@admin.sibptus.tomsk.ru> In-Reply-To: <20050917220415.GA73188@ratchet.nebcorp.com> References: <20050916073012.GA31056@admin.sibptus.tomsk.ru> <20050916225527.GT11689@ratchet.nebcorp.com> <20050917041700.GA46650@admin.sibptus.tomsk.ru> <20050917220415.GA73188@ratchet.nebcorp.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Danny Howard wrote: > > > > > > I have only ever mirrored disks with data on them. Its a question of > > > bootstrap - does the mirror comes first or does the data you are going > > > to mirror come first? > > > > Suppose the disk has valuable data in the last sector and you are > > going to create a mirror from this disk. What is going to happen when > > the last sector is overwritten with the mirror metadata? Your data will > > be lost, right? Suppose you need to access the last sector, access > > will be denied, right? > > >From what I have read, the gmirror stuff uses a sector of the disk that > isn't used at the end, and if you format a disk in such a way that that > free sector is unavailable, you can not set up a gmirror. Where have you read this? Where is it documented? I would like to read this too. I provided some screen output, do you mean to say I have formatted the disk incorrectly prior to mirroring it? > > If you find a scenario in which setting up a gmirror destroys data on a > disk, then please use send-pr to file a bug, so that that issue can be > fixed. I have found a scenario in which a gmirror prevents a partition from being newfs'ed. Is this enough for a PR ? > > But, if your question is "does gmirror do evil things to existing data" > then I'll answer that I and others have gmirrored quite a few disks with > existing data Have you and others tried to access the last sector of those disks? For example, with dd? May I ask you to try? > and I'm not aware that anybody has been bitten by your > hypothesis. Have you looked at the screen output I provided? What was I doing wrong? -- Victor Sudakov, VAS4-RIPE, VAS47-RIPN
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050918041618.GA59139>