Date: Fri, 3 Jul 2015 03:27:43 +0100 From: Ken Moffat <zarniwhoop@ntlworld.com> To: freebsd-questions@freebsd.org Subject: Re: slightly off topic: SMART error values for seagate drives Message-ID: <20150703022743.GA5293@milliways> In-Reply-To: <5595EE90.5050105@sneakertech.com> References: <5595EE90.5050105@sneakertech.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Jul 02, 2015 at 10:08:16PM -0400, Quartz wrote: > Is anyone familiar with exactly how the raw_read_error_rate, > reallocated_sector_count, seek_error_rate, hardware_crc_recovered, and > udma_crc_error_count values work for seagate drives? > > AFAIK at least some of these fields list the (average?) number of sectors > between errors, and thus a higher raw value is better ...I think. I have > several apparently healthy seagate drives with very high rrer/ser/her raw > values that seem to support this. > > I recently took an older disk out of storage, and as part of my system > building procedure I always run a few tests over it before putting it back > into service. Initially it also had high rrer/ser/her raw values. It failed > a SMART extended test at about 40% remaining with a read failure, and the > udma_crc_error_count jumped from 0 to 5. I know sometimes this can be just > transient flakiness, so I just zeroed out the entire drive with dd to > exercise all the sectors and force any remappings. Now, the > reallocated_sector_count bumped up to 9, and the raw_read_error_rate and > hardware_crc_recovered fields plummeted from the millions down to like 13 > ...and have since slowly risen to the 40's. > > I can't tell what's going on with SMART values anymore, every vendor does > things differently and nothing's ever documented. Is having the reallocated > sectors value go up still a bad thing, or did seagate change what this > means? Why did the read error rate and crc recovered fields bottom out, but > the seek error rate is still in the clouds? Is this drive failing, or fine, > or what? > Based on a small number of low-end consumer-level desktop machines, my experience is that an increasing reallocated sector count is a bad sign. In theory, we ought to be able to run with a number of reallocated sectors - but I only really notice when sectors become (temporarily) inaccessible. Generally, you get what you pay for - and the sorts of drives I buy are built down to a price. ĸen -- This one goes up to eleven!
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150703022743.GA5293>