Date: Fri, 11 May 2012 14:32:10 +0300 From: Alexander Motin <mav@FreeBSD.org> To: Karl Pielorz <kpielorz_lst@tdx.co.uk> Cc: freebsd-geom@FreeBSD.org Subject: Re: FreeBSD 9-R amd64 - graid, should it survive 'pulling' a disk? Message-ID: <4FACF8BA.4020500@FreeBSD.org> In-Reply-To: <61CDA17B687F1C733EE1B017@OctaHexa64-MkII> References: <4FABDE10.8090304@FreeBSD.org> <61CDA17B687F1C733EE1B017@OctaHexa64-MkII>
next in thread | previous in thread | raw e-mail | index | archive | help
On 11.05.2012 14:08, Karl Pielorz wrote: > --On 10 May 2012 18:26 +0300 Alexander Motin <mav@FreeBSD.org> wrote: >> This panic is not in graid code, but it can be called a problem of the >> graid's RAID1 implementation. If some disk returns _write_ failure, that >> failure may now be reported to higher levels. Problem is that UFS SU code >> panics on these errors in some cases. I'll try to look on it nearest >> time. > > So for clarity - what you're saying is if one of the servers drops a > disk from it's graid RAID1 array (for whatever reason) while writing, > the error could 'bubble up' and have the UFS code panic, because of it? > (even though the write would have completed on the other drive)? Yes, sometimes bubble and in rare cases panic. I am working on the first part right now. > Is the same true for reads? No, On read error graid repeats reading from another disk and does remapping write to the original one. Error returned only if all reads have failed. -- Alexander Motin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4FACF8BA.4020500>