From: Justin P. <jp...@lu...> - 2006-09-19 12:46:40
|
On Tue, 19 Sep 2006, G. Vincent Castellano wrote: > After having two large disks in my home server go bad, I am finally motivated > to start setting up smartd there and at my day job. It's working out very > well so far except for a problem with my newest disk: > > Seagate Barracuda 500GB - 7200rpm - 100MBps Ultra ATA/100 (ATA-6) > > The smartd.conf looks contains a single line: > > DEVICESCAN -a -m gvc -M test -s (L/../../7/01|S/../../(1|2|3|4|5|6)/01) > > This started a long test as expected Sunday morning (2006-09-17) but over two > days later smartctl -a showed it 90% complete. I might not have noticed it > but for the fact that the device throughput was terrible (I don't have the > exact numbers, but I think it was only allowing around a megabit per second). > When I start the test from the command line it estimates 255 runtime. > > I aborted the long test and performance returned to normal. This disk is in > constant use as a repository for a (high-quality) MP3 stream, averaging > around 180 Mbps. > > Output of smartctl -a on the device is attached. > > Here are some entries from the smartd syslog output, I would be interested in > finding out how I can learn more about how to interpret these messages, even > if they don't indicate anything related to the long-test performance problem. > > Sep 17 04:19:51 tuvok smartd[23334]: Device: /dev/hdb, SMART Usage Attribute: > 195 Hardware_ECC_Recovered changed from 49 to 50 > Sep 17 05:19:49 tuvok smartd[23334]: Device: /dev/hdb, SMART Usage Attribute: > 190 Unknown_Attribute changed from 49 to 50 > Sep 17 06:19:49 tuvok smartd[23334]: Device: /dev/hdb, SMART Prefailure > Attribute: 7 Seek_Error_Rate changed from 68 to 69 > > --gvc > I noticed this as well on the Seagate 16MB/7200RPM/SATAII drives, I stopped long testing and do short tests only on those drives. Justin. |