Hi Eugenix
You raised an very interesting question.
I set up my home scenario for a backup of notebook data and, of course, other large data storage.
I got curious about your enquiry and found that my MBWE, 2 months old, now has this value at 72757.
It took my sunday morning researching and trying to understand and figure out if is there any workaround.
Although WDC states that the drives are rated for > 1 million Load cycles (see article at http://wdc.custhelp.com/cgi-bin/wdc.cfg/php/enduser/std_adp.php?p_faqid=5357 it's pretty interesting), I feel myself very uncomfortable for nearing 10% of lifespan in only 2 months of usage.
I tried the hdparm -B 255 /dev/sda to turn the head parking off, but I get this error:
[root@DadosCasa init.d]# hdparm -B 255 /dev/sda
/dev/sda:
setting Advanced Power Management level to disabled
HDIO_DRIVE_CMD failed: Input/output error
APM_level = not supported
I even disassembled my hard drive case (ooh boy, so many broken latches…) to try to tune the hdparm on my other machine, then I found that I have no SATA cable over here (pissed)
Then I took other line of investigation. I coupled strace + vmstat (actually, dstat) to figure out what's writing or polling data on the hard drive (used strace -e trace=file -p <pid>) and I found some activity over here:
Cron job polling for content change:
[pid 3184] stat64("crontabs", {st_mode=S_IFDIR|0755, st_size=17592186044416, ...}) = 0
[pid 3184] stat64("/opt/etc/cron.d", {st_mode=S_IFDIR|0755, st_size=17592186044416, ...}) = 0
[pid 3184] stat64("/opt/etc/crontab", {st_mode=S_IFREG|0600, st_size=17592186044416, ...}) = 0
Samba (my wife on her station writing her scientific papers) polling for the directory and runtime data:
[pid 4195] lstat64("Dados", {st_mode=S_IFDIR|0775, st_size=17592186044416, ...}) = 0
[pid 4195] stat64("Dados/monografia", {st_mode=S_IFDIR|0775, st_size=17592186044416, ...}) = 0
[pid 4195] open("Dados", O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY) = 25
[pid 4195] fstat64(25, {st_mode=S_IFDIR|0775, st_size=17592186044416, ...}) = 0
[pid 4195] stat64("", 0xbe8be790) = -1 ENOENT (No such file or directory)
[pid 4195] open("/var/log/samba/log.smbd", O_WRONLY|O_CREAT|O_APPEND|O_LARGEFILE, 0644) = 25
[pid 4195] stat64("/usr/private/user_smb_conf/.overall_share", {st_mode=S_IFREG|0600, st_size=17592186044416, ...}) = 0
[pid 4195] stat64("/etc/samba/smb.conf", {st_mode=S_IFREG|0600, st_size=17592186044416, ...}) = 0
[pid 4195] fstat64(25, {st_mode=S_IFREG|0644, st_size=17592186044416, ...}) = 0
[...]
And /usr/sbin/wixhwmonitor checking from time to time how things are going.
Well, what's the point.
1. Writing a document on Word or PowerPoint does read/write something sporadically in the drive; then triggering the load cycle.
2. For people which usage behavior reflects item 1, it may become a concern.
3. By using hdparm, we can't disable the power management (don't know if it's a disk or OS limitation)
4. If I manage to 'hdparm' the drive on other computer, will it persist across a power on-off?
5. If not, definitely it should be an RFE on MBWE. There's already an spindle off disable configuration on MBWE; it should be also coupled to -B 255. But we have to overcome the problem that I wrote in the beginning of this post.
6. Anyone? Good/bright/insightful ideas?
Best,
- RF.
