Got new HDD for 110€. Its a Samsung/Seagate ST2000LM003 HN-M201RAD. It is also known as Samsung Spinpoint M9T. Samsung labels this device as MOBILE SATA Drive. It seems those drives are (software?) adjusted to park their read/write head much more often than in an conventional „desktop“ drive to prevent shock errors while moving laptops/notebooks. The datasheet states the drive would live longer than 600000 Controlled Ramp Load/Unload. Here is how I calculate how long my drive should be alive. Get the values by:
smartctl –all /dev/sda | grep -e „Hour“ -e „Load_Cycle“
9 Power_On_Hours 0x0032 100 100 000 Old_age Always – 11
225 Load_Cycle_Count 0x0032 100 100 000 Old_age Always – 297
297/11=27 cycles per hour (ST2000LM003). I have another 2.5 HDD that is 4 years old that only has 5.7 cycles per hour. 27*24*365=236520 cycles per year. Now divide the estimated 600000 by 236520 cycles per year and I get 2.5 years lifespan for the drive which is quite short. Next step is to find out if this cylce count can be lowered somehow, for example with hdparm. But my disk already shows: smartctl -s on –xall /dev/sda AAM feature is: Disabled APM feature is: Disabled and hdparm -B /dev/sda /dev/sda: APM_level = off I will gather some more Cycle Count values over the days to get a more average value. Then I will try to set a value via hdparm.
- UPDATE: overnight cycle count steady inceased to 737 for 31 hours usage. That gives 2.8 years lifespan.
- UPDATE: I was able to lower the cycle count to 5 per hour. I used „hdparm -B 254 /dev/sda“ to enable APM on the disk and „hdparm -S 120 /dev/sda“ to set spindown time to 10 minutes.
- UPDATE: After six months my load cycle count seems to stall at 4/hour. No increase of any other error values. Two things to mention: one may need to reset the two hdparm commands after every reboot! Otherwise the hdd is reset to factory default apm values. I used a start script in linux for that job. Sometimes I can still hear some strange short whistle from the disk. I dont know whats the cause of this but since I dont see error values changes I think thats just some weird internal disk function.