I've been following the current controversy about whether Ubuntu kills laptop drives: https://launchpad.net/bug59695.html I don't use Ubuntu but I did smartctl -d ata -a /dev/sda | grep 193 and got a count of over 1 million. The product manual for my drive (Seagate) says 300K on/off cycles at 25 degrees, 50% humidity and only 100K in less favorable conditions. Can I believe the output of smartctl? I've now done all the things needed to turn off agressive power-saving but I'm wondering whether I should get a new drive ASAP. Is there some way to tell whether the drive is damaged? Bob T.

Bob Tennent <BobT@cs.queensu.ca> writes: >I've been following the current controversy about whether Ubuntu kills >laptop drives: > >https://launchpad.net/bug59695.html > >I don't use Ubuntu but I did > > smartctl -d ata -a /dev/sda | grep 193 > >and got a count of over 1 million. The product manual for my drive >(Seagate) says 300K on/off cycles at 25 degrees, 50% humidity and only >100K in less favorable conditions. Can I believe the output of smartctl? Yes. I have heard the load cycles on my laptop (see <2005Dec24.121658@mips.complang.tuwien.ac.at> ff.) on an earlier disk. >I've now done all the things needed to turn off agressive power-saving >but I'm wondering whether I should get a new drive ASAP. Is there some >way to tell whether the drive is damaged? The first thing is to look at the smartctl output. E.g., on my current drive it says: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 193 Load_Cycle_Count 0x0032 089 089 000 Old_age Always - 237320 As you can see, in my case the VALUE has not reached the THRESHOLD yet (VALUE typically starts at 100 or 200 and gets smaller over time). If it has not reached THRESHOLD yet on your drive, your drive is fine in theory. Even if it has reached the THRESHOLD, I would not expect it to fail soon. So, just wait, and if it fails, put in a new hard disk and put your backup on it (you do have a backup, right?). On the earlier disk, I turned off the frequent load cycles with hdparm -B254 /dev/hda I have read some warnings about using that, so use at your own risk. - anton -- M. Anton Ertl Some things have to be seen to be believed anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html

On Wed, 31 Oct 2007 22:50:58 GMT, Anton Ertl wrote: > Bob Tennent <BobT@cs.queensu.ca> writes: >> >>I did >> >> smartctl -d ata -a /dev/sda | grep 193 >> >>and got a count of over 1 million. > > The first thing is to look at the smartctl output. E.g., on my > current drive it says: > > ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE > 193 Load_Cycle_Count 0x0032 089 089 000 Old_age Always - 237320 > > As you can see, in my case the VALUE has not reached the THRESHOLD yet > (VALUE typically starts at 100 or 200 and gets smaller over time). If > it has not reached THRESHOLD yet on your drive, your drive is fine in > theory. Even if it has reached the THRESHOLD, I would not expect it > to fail soon. My smartctl output is 193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 1032808 and I thought the value at the right was the Load_Cycle_Count. What is a "RAW_VALUE". So my VALUE is just above the threshold of 000; not encouraging. Yes, I have a backup. How can VALUES/RAW_VALUES be related to the maximum load/unload cycles of the product manual? Bob T.

Bob Tennent <BobT@cs.queensu.ca> writes: >On Wed, 31 Oct 2007 22:50:58 GMT, Anton Ertl wrote: > > The first thing is to look at the smartctl output. E.g., on my > > current drive it says: > > > > ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE > > 193 Load_Cycle_Count 0x0032 089 089 000 Old_age Always - 237320 > > > > As you can see, in my case the VALUE has not reached the THRESHOLD yet > > (VALUE typically starts at 100 or 200 and gets smaller over time). If > > it has not reached THRESHOLD yet on your drive, your drive is fine in > > theory. Even if it has reached the THRESHOLD, I would not expect it > > to fail soon. > >My smartctl output is > >193 Load_Cycle_Count 0x0032 001 001 000 Old_age Always - 1032808 > >and I thought the value at the right was the Load_Cycle_Count. What >is a "RAW_VALUE". In this case, it is probably the number of load/unload cycles. > So my VALUE is just above the threshold of 000; not >encouraging. I would not worry about it too much. If the drive can stand one million load cycles, it probably won't break when you do another million. > Yes, I have a backup. How can VALUES/RAW_VALUES be related >to the maximum load/unload cycles of the product manual? If RAW_VALUES gives the number of cycles, then VALUE should reach THRESH when RAW_VALUES reaches the maximum number given in the product manual. - anton -- M. Anton Ertl Some things have to be seen to be believed anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html

Bob Tennent <BobT@cs.queensu.ca> writes: >On Thu, 01 Nov 2007 11:58:34 +0100, Pascal Hambourg wrote: >Here's the entry: > > Load/Unload (U/UL) cycles > > 25°C, 50% relative humidity 300,000 software-controlled power on/off cycles > 20,000 hard power on/off cycles > >This is from > >http://www.seagate.com/support/disc/manuals/ata/momentus_pm.pdf > >If that's not the relevant spec, what is? That manual does not say AFAICS. The SMART attributes having to do with power are 12 Power_Cycle_Count and possibly 4 Start_Stop_Count The latter may or may not be what is meant with "software-controlled power on/off cycles" A load cycle is when the drive moves the head arms into the parking position and back into the reading position. Laptop drives often move the heads to the parking position, because then they are better protected against shock. Under Linux moving the heads to the parking position is pretty pointless, because a few seconds later Linux unparks the heads (probably to write atimes or something). Laptop mode should cure this, but I have not gotten around to using it on mine. - anton -- M. Anton Ertl Some things have to be seen to be believed anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html

On Thu, 01 Nov 2007 13:09:38 GMT, Anton Ertl wrote: > A load cycle is when the drive moves the head arms into the parking > position and back into the reading position. Laptop drives often move > the heads to the parking position, because then they are better > protected against shock. Under Linux moving the heads to the parking > position is pretty pointless, because a few seconds later Linux > unparks the heads (probably to write atimes or something). Laptop > mode should cure this, but I have not gotten around to using it on > mine. So moving to the parking position when the drive is not in use is good, provided that the partitions are mounted with noatime? Why has the discussion at the Ubuntu site focussed on hdparm -B settings? Bob T.

Bob Tennent <BobT@cs.queensu.ca> writes: >On Thu, 01 Nov 2007 13:09:38 GMT, Anton Ertl wrote: > > > A load cycle is when the drive moves the head arms into the parking > > position and back into the reading position. Laptop drives often move > > the heads to the parking position, because then they are better > > protected against shock. Under Linux moving the heads to the parking > > position is pretty pointless, because a few seconds later Linux > > unparks the heads (probably to write atimes or something). Laptop > > mode should cure this, but I have not gotten around to using it on > > mine. > >So moving to the parking position when the drive is not in use is good, >provided that the partitions are mounted with noatime? Maybe, I have not tried it. Wasn't there some work on lazy atime writing? And in any case, with laptop mode you would not just park the heads, but also spin down the drive for some time, which would be even better than just parking (less power). > Why has the >discussion at the Ubuntu site focussed on hdparm -B settings? I have not read that, so I cannot say. - anton -- M. Anton Ertl Some things have to be seen to be believed anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html