Hard drive will not stay in standby mode

[Greg Michalec]

Binary package hint: hdparm

Details:
I just got a new SATA hard drive and moved all my stuff to it, but I'm keeping my old PATA around for backups, media, etc. However, compared to the new one, it is noisy! I'd like for it to stay asleep/spun down as much as possible, but I'm having problems making it do so. When I issue the command "hdparm -y /dev/sda" it spins down, but it inevitably spins back up again after anywhere from 5 to 60 seconds. Also, 'hdparm -S 1 /dev/sda' seems to have no effect.

I figured that there was some application polling the data off the drive, so I tried unmounting its partitions. Even with no unmounted partitions, the drive still continues to spin up! I know i've done this in the past with older linux's. Maybe something to do with the new libata driver?

Thanks

[Greg Michalec (greg-primate)]

This problem is related to hddtemp, not hdparm. See Bug #160621

[kernelOfTruth (dalinuxlova)]

last time I used my system I saw that this was also related / caused by:

- hddtemp
- sensors-applet
- smartmontools

(I'll add more if I encounter them)

[Antti Kaihola (akaihola)]

I'm pretty sure I also had disks spin up after a couple of hours on a freshly installed Mythbuntu 9.10 Karmic Alpha 6 box. I don't have hddtemp, sensors-applet nor smartmontools installed.

I'll get back with more accurate information.

[Antti Kaihola (akaihola)]

Ok, this is embarrassing. I had a partition on a spun-down disk being used as swap space.

[Antti Kaihola (akaihola)]

After eliminating swap my /dev/sdb still spins up about once an hour.

I don't have any of the following installed:
- hddtemp
- sensors-applet
- smartmontools

In "cat /proc/mounts" I don't see any references to /dev/hdb or UUIDs of any of its partitions.

I tried monitoring disk access and got no indication of any process accessing /dev/sdb* when the disk spun up or at any other time. I used this script to monitor:
    echo 1 >/proc/sys/vm/block_dump
    while ( ! dmesg -c|grep sdb ); do sleep 1; done
    echo 0 >/proc/sys/vm/block_dump

Is there any way I could "freeze" the kernel after issuing the hdparm -S/y/Y command so that I can be sure neither the kernel nor any process is able to access the disk? That way I could make sure it's not the disk itself or the controller which decides to spin up the disk.

[Antti Kaihola (akaihola)]

Ok, my problem is different since my spin-up interval is exactly 30 minutes. See Bug #435190