ZFS Special device - Spin down

I’ve been investigating why my hdd don’t spin down - Some months ago I added ZFS special devices and I have a filesystem where the special block size is equal to the record size

It is my understanding that would make everything on it written to the fast nvme special device.

However; I have noticed recently that my disks stopped spinning down once adding this, and running this;

dstat -tdD /dev/sda --top-io

07-04 18:56:54|  28M   18M|cop          80M   80M
07-04 18:56:55|  26M 8800k|cop         120M  120M
07-04 18:56:56|  29M   23M|cop          96M   96M
07-04 18:56:57|  20M   50M|cop          72M   72M
07-04 18:56:58|  23M   60M|cop          84M   84M
07-04 18:56:59|  28M   44M|cop         132M  132M
----system---- --dsk/sda-- ----most-expensive----
     time     | read  writ|     i/o process      
07-04 18:57:00|  27M   54M|cop         100M  100M
07-04 18:57:01|  11M   22M|cop          80M   80M

Seems to suggest that device sda (Spinning rust) is being read/written. This was me copying a large MP4 to my ‘fast’ filesystem

Any help would be appreciated

For anyone reading this in the future; I was not able to stop my drives spinning up using a zpool that assigned all blocks as special blocks. I never found out why that was the case.

I instead changed my setup, to create separate pools, based on partitions and was able to achieve what I wanted using this setup.

Pools now look like this

  pool: Fast
 state: ONLINE
  scan: resilvered 75.3G in 00:04:05 with 0 errors on Thu Apr 18 13:52:47 2024
config:

        NAME                                                    STATE     READ WRITE CKSUM
        Fast                                                    ONLINE       0     0     0
          mirror-0                                              ONLINE       0     0     0
            nvme-Samsung_SSD_990_PRO_1TB_S6Z1NJ0W333776L-part2  ONLINE       0     0     0
            nvme-Samsung_SSD_990_PRO_1TB_S6Z1NF0WB12620F-part2  ONLINE       0     0     0
        logs
          nvme-INTEL_SSDPE21D280GA_PHM2749000VL280AGN-part1     ONLINE       0     0     0

errors: No known data errors

  pool: NAS
 state: ONLINE
  scan: resilvered 15.3G in 00:00:28 with 0 errors on Thu Apr 18 00:27:21 2024
config:

        NAME                                                    STATE     READ WRITE CKSUM
        NAS                                                     ONLINE       0     0     0
          raidz1-0                                              ONLINE       0     0     0
            ata-WDC_WD10EURX-63C57Y0_WD-WCC4JCC6KRU9            ONLINE       0     0     0
            ata-WDC_WD10EURX-63C57Y0_WD-WCC4JL4W41VE            ONLINE       0     0     0
            ata-WDC_WD10EURX-83UY4Y0_WD-WCC4J3LCTDFA            ONLINE       0     0     0
            ata-WDC_WD10EZRX-00A3KB0_WD-WCC4J3EXK40P            ONLINE       0     0     0
        special
          mirror-1                                              ONLINE       0     0     0
            nvme-Samsung_SSD_990_PRO_1TB_S6Z1NF0WB12620F-part1  ONLINE       0     0     0
            nvme-Samsung_SSD_990_PRO_1TB_S6Z1NJ0W333776L-part1  ONLINE       0     0     0
        logs
          nvme-INTEL_SSDPE21D280GA_PHM2749000VL280AGN-part2     ONLINE       0     0     0

errors: No known data errors

ZFS and spin-down doesnt work. just accept it.

It definitely is more work; but I am down to around 22h a day non-spinning

Depends on what is being accessed, obviously

What command(s) do you use to get your drives to spin down?

I use hd-idle - And I get the git version, the debian one didn’t work particularly well