Spindowns, optimisations

[spamalam]

Hi,

 

I'm trying to reduce heat, power and optimise the life of my drives and wondering if anyone could review this setup/provide advice.  I have a mixture of drives, brands and live in a climate that's not so great for cooling (South of Spain).  I will invest in better cooling long term but noise is a big problem and its a constant struggle to balance.  One thing I want to do is avoid unnecessary spinning disks, for this reason I run dockers outside of the array, try and use cache and only do an overnight array backup.  My expectation is that the only time disks should be spinning is:

1. Parity actions - rebuild, check, etc.
2. Active read from a share, with a power down
3. Radarr/Sonarr/Plex Library scans, which are scheduled for nightly (I hope)

I've written some scripts to try and do things like save my moving listing to a txt file and read from that instead of listing out the share actively.  I tried to use cache folders plugin but I'm not sure its working well.

 

I expect that most of the time, disks should be idle, powered down and waiting for a command.

 

I have the following configuration for the array:

2x Parity

20x Data Disks (JBOD, mix of drives)

2x Backup Data Disks (in case a drive dies or i run out of space)

1x Cache Disk

2x SSD outside of array (for appdata, etc.)

 

Recently I have noticed that many disks are spinning 'constantly', so I wanted to investigate what's going on.  The weird one is that the two hot-backup data disks are the ones spinning most of the time and they really should not be as they are never used other than display information in the unassigned drives plugin.

 

Unfortunately I'm in the middle of a parity rebuild which means drives are spinning a lot, but I left everything untouched overnight, and ran a quick hdparm check this morning:

root@beyonder-nas:/boot/custom/scripts# /bin/bash check.sh
/dev/disk/by-id/ata-CT2000MX500SSD1_1923E209C6B6: drive state is: active/idle
/dev/disk/by-id/ata-HGST_HDN724040ALE640_PK2381PBG1X9LT: drive state is: standby
/dev/disk/by-id/ata-ST12000VN0007-2GS116_ZCH02PFF: drive state is: active/idle
/dev/disk/by-id/ata-ST12000VN0007-2GS116_ZCH07SCG: drive state is: active/idle
/dev/disk/by-id/ata-ST4000VN000-1H4168_Z305P6DR: drive state is: standby
/dev/disk/by-id/ata-ST8000AS0002-1NA17Z_Z840FN16: drive state is: standby
/dev/disk/by-id/ata-Samsung_SSD_840_EVO_250GB_S1DBNSAFB35169F: drive state is: active/idle
/dev/disk/by-id/ata-Samsung_SSD_850_EVO_250GB_S2R6NX1J916883B: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD100EMAZ-00WJTA0_1EGP429N: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD120EDAZ-11F3RA0_8CK5TY6E: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD120EDAZ-11F3RA0_8CKBEPJE: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD140EMFZ-11A0WA0_9RHMRJPC: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD140EMFZ-11A0WA0_9RJ6ESGC: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD140EMFZ-11A0WA0_9RJ7WTJC: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD140EMFZ-11A0WA0_Z2GBNUWT: drive state is: active/idle
/dev/disk/by-id/ata-WDC_WD40EFRX-68WT0N0_WD-WCC4E1047137: drive state is: standby
/dev/disk/by-id/ata-WDC_WD40EFRX-68WT0N0_WD-WCC4E1058061: drive state is: standby
/dev/disk/by-id/ata-WDC_WD40EFRX-68WT0N0_WD-WCC4ELEKNUXP: drive state is: standby
ARECA for /dev/disk/by-id/scsi-HDN724040ALE640_PK2381PBJ39RWT
/dev/disk/by-id/scsi-SATA_CT2000MX500SSD1_1923E209C6B6: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_HGST_HDN724040AL_PK2381PBG1X9LT: drive state is: standby
/dev/disk/by-id/scsi-SATA_ST12000VN0007-2G_ZCH02PFF: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_ST12000VN0007-2G_ZCH07SCG: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_ST4000VN000-1H41_Z305P6DR: drive state is: standby
/dev/disk/by-id/scsi-SATA_ST8000AS0002-1NA_Z840FN16: drive state is: standby
/dev/disk/by-id/scsi-SATA_Samsung_SSD_840_S1DBNSAFB35169F: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_Samsung_SSD_850_S2R6NX1J916883B: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD100EMAZ-001EGP429N: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD120EDAZ-118CK5TY6E: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD120EDAZ-118CKBEPJE: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD140EMFZ-119RHMRJPC: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD140EMFZ-119RJ6ESGC: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD140EMFZ-119RJ7WTJC: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD140EMFZ-11Z2GBNUWT: drive state is: active/idle
/dev/disk/by-id/scsi-SATA_WDC_WD40EFRX-68W_WD-WCC4E1047137: drive state is: standby
/dev/disk/by-id/scsi-SATA_WDC_WD40EFRX-68W_WD-WCC4E1058061: drive state is: standby
/dev/disk/by-id/scsi-SATA_WDC_WD40EFRX-68W_WD-WCC4ELEKNUXP: drive state is: standby
ARECA for /dev/disk/by-id/scsi-ST12000VN0007-2G_ZCH07BKV
ARECA for /dev/disk/by-id/scsi-ST4000VN000-1H41_S300YVM0
ARECA for /dev/disk/by-id/scsi-ST4000VN000-1H41_Z300N10C
ARECA for /dev/disk/by-id/scsi-ST8000AS0002-1NA_Z840A661
ARECA for /dev/disk/by-id/scsi-ST8000AS0002-1NA_Z840A7TZ
ARECA for /dev/disk/by-id/scsi-WD100EMAZ-00WJTA_1EGL148N
ARECA for /dev/disk/by-id/scsi-WD120EDAZ-11F3RA_8CKBJWUL

 

Let's pick on the hot backup disk here:

/dev/disk/by-id/scsi-SATA_WDC_WD140EMFZ-11Z2GBNUWT: drive state is: active/idle

Shouldn't this be in standby if its outside the array and unused?  If i issue:

 hdparm -y /dev/sdv

it correctly goes to standby.  How do i prevent it spinning up whenever unraid does something?

 

Some things I've noticed:

1. By default every diskx was exposed as a share, I don't recall ever setting this... would this have caused disks to be spinning?  I've disabled all smb/nfs now, but shouldn't this be disabled by default (security/performance/etc)?  I never setup those shares explicitly.
2. appdata was set to cache prefer, but shouldn't this be only?  I don't think I defined this share and it was created when docker was introduced, was it a victim of unraid upgrades?  I've set this to only and deleted the appdata folder from all the drives.
3. If I hit the "Main" tab in the UI, it seems to spin-up all the disks and that the spin-up/spin-down button is negated if i hit F5.  Is this normal?  Can it be prevented?  It seems like a bug to do this, particularly for disks UNRAID isn't using this disk for anything?  Or is the unassigned plugin the culprit?  For now I've switched my home tab to dashboard.
4. ARECA doesn't give me valid feedback on disk states, but the raid controller is managing idle times.  Unraid is set to 15min idle time, whilst areca is set to 30min.  Would unraid try and mess with these or will it live nicely with this setting ?

 

Any help / advice would be appreciated!

 

I'm attaching a diagnostic in case anything is faulty/misconfigured.  I'm in the middle of a parity build unfortunately so there's more disk activity than usual.

 

beyonder-nas-diagnostics-20200304-1318.zip

[spamalam]

Parity check is finished, and the unassigned drive /dev/sdv is still spinning and I can't figure out why?

 

 

Attaching diagnostics from post-parity rebuild.

beyonder-nas-diagnostics-20200305-1149.zip

[bonienl]

Unassigned devices are not included in the spin-down control.

[spamalam]

On 3/5/2020 at 1:51 PM, bonienl said:

Unassigned devices are not included in the spin-down control.

Okay that explains it, is there a way to get them to be?  I managed to keep it powered down by not hitting the /Main page, but when I do it spins them up.  Is unraid doing a /dev/* action that's causing the spin-ups?

Edited March 7, 2020 by spamalam

[bonienl]

1 hour ago, spamalam said:

Okay that explains it, is there a way to get them to be?

The feature "multiple pools" which allows to create more pools besides the cache pool, is currently under development.

Disks in a user defined pool will have spin up/down logic.

 

You need to wait though 😁