Unshared disk don't sleep

lacc47
lacc47 Posts: 4  Freshman Member
edited March 2019 in Personal Cloud Storage
I have nas326 with two disks. I configured two independent volumens on the disks, so there is no RAID.
One disk (volume1) is for daily use, and is shared to internal network. Another disk (volume2) is for making regulary backups from volume1. Volume2 has no shared folder on network, only NAS can access it (while making backup). I see on the led's, that if I use disk1, then disk2 is also awake from sleeping mode, and then it enters only together with disk1 to sleeping mode. I also see on leds, that disk2 sometimes work with something... but is not even shared!

I wish to keep disk2 always sleeping and unaccessed, exept when backup from disk1 is running.
thanks
lacc47

#NAS_Mar_2019

Accepted Solution

  • Mijzelf
    Mijzelf Posts: 2,598  Guru Member
    First Anniversary 10 Comments Friend Collector First Answer
    Answer ✓
    I got a log. It contains a lot of samba access to md2:
    <div>smbd(19842): READ block 745223424 on md2 (1024 sectors)<br></div>
    Samba is the server which is responsible for 'windows' shares, and md2 is the first data volume. So this is normal behavior, when accessing a share.

    Further there is access to loop0 and md0:
    httpd(19950): READ block 162346 on loop0 (8 sectors)<br>loop0(1550): READ block 735760 on md0 (16 sectors)<br>
    Here the webserver (httpd) is accessing loop0. Loop0 is a loopdevice, which is a file which behaves as a partition. On this NAS the GUI part (and more) of the firmware is located in a filesystem on that loopdevice. So by accessing the webinterface you are also accessing loop0. The file backing loop0 is on md0, and so accessing loop0 will lead to access of md0, which is a raid1 array, dedicated for that backing file, and which spans all disks. So by accessing the webinterface your 2nd disk is accessed.
    Theoretically you can remove the 2nd disk from the md0 array (and also from md1, which is a raid1 containing the swap), but I don't know if the firmware will put it back, and if the firmware will still recognize that 2nd disk.
    You can also try to set 'write mostly' for the md0 and md1 members on disk 2. As the loop device is read only, there will be no writing to that raid member, and reading is as far as possible disabled. Maybe that will keep the disk from spinning up. That is not true for the swap, of course. But I don't think swap will be used a lot.
    A third option is to use Tweaks to disable as many unneeded daemons as possible. When you are only accessing the shares on disk1, there is no need to read loop0. But some daemons (for instance fileye.pyc) use files from loop0.



All Replies

  • Mijzelf
    Mijzelf Posts: 2,598  Guru Member
    First Anniversary 10 Comments Friend Collector First Answer
    The 3rth party package 'Tweaks', which can be installed using 'MetaRepository' has a tool DiskMonitor, which in most cases can tell what is accessing the disk.
  • Mijzelf
    Mijzelf Posts: 2,598  Guru Member
    First Anniversary 10 Comments Friend Collector First Answer
    Answer ✓
    I got a log. It contains a lot of samba access to md2:
    <div>smbd(19842): READ block 745223424 on md2 (1024 sectors)<br></div>
    Samba is the server which is responsible for 'windows' shares, and md2 is the first data volume. So this is normal behavior, when accessing a share.

    Further there is access to loop0 and md0:
    httpd(19950): READ block 162346 on loop0 (8 sectors)<br>loop0(1550): READ block 735760 on md0 (16 sectors)<br>
    Here the webserver (httpd) is accessing loop0. Loop0 is a loopdevice, which is a file which behaves as a partition. On this NAS the GUI part (and more) of the firmware is located in a filesystem on that loopdevice. So by accessing the webinterface you are also accessing loop0. The file backing loop0 is on md0, and so accessing loop0 will lead to access of md0, which is a raid1 array, dedicated for that backing file, and which spans all disks. So by accessing the webinterface your 2nd disk is accessed.
    Theoretically you can remove the 2nd disk from the md0 array (and also from md1, which is a raid1 containing the swap), but I don't know if the firmware will put it back, and if the firmware will still recognize that 2nd disk.
    You can also try to set 'write mostly' for the md0 and md1 members on disk 2. As the loop device is read only, there will be no writing to that raid member, and reading is as far as possible disabled. Maybe that will keep the disk from spinning up. That is not true for the swap, of course. But I don't think swap will be used a lot.
    A third option is to use Tweaks to disable as many unneeded daemons as possible. When you are only accessing the shares on disk1, there is no need to read loop0. But some daemons (for instance fileye.pyc) use files from loop0.



Consumer Product Help Center