Date: Thu, 23 Jul 2009 15:42:47 -0500 (CDT) From: "Sean C. Farley" <scf@FreeBSD.org> To: Bruce Simpson <bms@incunabulum.net> Cc: freebsd-stable@FreeBSD.org Subject: Re: ataraid's revenge! (Was: Re: A nasty ataraid experience.) Message-ID: <alpine.BSF.2.00.0907231536350.78159@thor.farley.org> In-Reply-To: <4A68B2A0.8050509@incunabulum.net> References: <200901232244.n0NMiRmM098646@lurza.secnetix.de> <46acbb3e-71bc-4cff-93d7-59b48a1a9302@exchange01.ecp.noc> <4A68B2A0.8050509@incunabulum.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 23 Jul 2009, Bruce Simpson wrote: > 6 months on, ataraid(4) did it again. > > This time, I was lucky -- I caught in in time, but the damage to the > filesystem meant having to use fsdb to NULL out the affected inodes; > mounting read-only, tarring, and untarring across the network, after a > newfs, let me save the affected partition. > All I was doing at the time was srm'ing a few sensitive files; all > the processes wedged in WCHAN getblk. It seems ataraid(4) is not > robust against temporary drive/controller problems. The SMART logs on > the affected array drives all check out just fine, there are no bad > block remaps. > > So, time to either buy a hardware RAID controller, or move to ZFS... Out of fear of what ataraid may do to me especially with Intel MatrixRAID[1], I switched to using gmirror awhile back, and it has worked well. When I buy some new drives soon, I am considering using gvinum with 8.0, but I need to find out more. Anyone know if I can boot off of a gvinum partition and/or how it works (or does not) with various label schemes? Sean 1. http://wiki.freebsd.org/JeremyChadwick/ATA_issues_and_troubleshooting -- scf@FreeBSD.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.00.0907231536350.78159>