From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andreas Klauer Subject: Re: recovering failed raid5 Date: Sat, 29 Oct 2016 14:02:30 +0200 Message-ID: <20161029120230.GA4725@metamorpher.de> References: <20161027160400.GA21042@metamorpher.de> <715b259f-1e56-9606-edc4-3e5c4d57744b@shenkin.org> <20161028133304.GA11564@metamorpher.de> <20161029152951.62add3ca@natsu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20161029152951.62add3ca@natsu> Sender: linux-raid-owner@vger.kernel.org To: Roman Mamedov Cc: linux-raid@vger.kernel.org List-Id: linux-raid.ids On Sat, Oct 29, 2016 at 03:29:51PM +0500, Roman Mamedov wrote: > And if there's an unreadable sector on rebuild as a drive found its 8th bad > sector after 3 more years of perfect operation, that's not a problem either, > because the setup they run in, is RAID6. But if such disks are acceptable to run in a RAID, and you advertize it as such, you have to expect to see RAIDs where every single disk has a dozen reallocated sectors and a history of read errors to go with it. Is that still fine? Do you expect to be lucky every time? RAID-6 is not magic, either. Sooner or later, it will fail, too. Keep ignoring errors in RAID-5 and you'll see double failure. Keep ignoring errors in RAID-6 long enough and you'll see triple failure. All disks fail and many of them do silently, undetected if untested. If you rented a server in a datacenter, thus entitled to working hardware, would you create a ticket on read failure or not? Regards Andreas Klauer