Might only be a problem with my system, but thought I'd mention it.

 

unRAID Server Pro 5.0-rc6-r8168-test2 - no addons.

System specs: AMD Athlon II X2, Asus M5A97 EVO, 4gb ram, 6 hdd's all running from onboard Sata 3 controller.

Sata 1 - Disk 1 - 2TB WD Green - TV Shows (A-G)

Sata 2 - Disk 2 - 2TB WD Green - TV Shows (H-R)

Sata 3 - Disk 3 - 2TB WD Green - TV Shows (S-Z)

Sata 4 - Disk 4 - 2TB WD Green - Movies

Sata 5 - Disk 5 - 2TB Hitachi - Parity

Sata 6 - Disk 6 - 2TB Hitachi - Cache

 

Since upgrading, was running 5.0-rc5-r8168, every time the mover runs, via script and manually, it spins up Disk 1, even if nothing is being written to Disk 1.

 

All disks, but the cache drive, where spun down when I run the mover manually.

 

Sep 13 12:06:15 Server emhttp: shcmd (80): /usr/local/sbin/mover |& logger &

Sep 13 12:06:15 Server logger: mover started

Sep 13 12:06:15 Server logger: moving Movies/

Sep 13 12:06:15 Server logger: ./Movies/Action/Saving Private Ryan.mkv

Sep 13 12:06:15 Server logger: .d..t...... ./

Sep 13 12:06:15 Server logger: .d..t...... Movies/

Sep 13 12:06:15 Server logger: .d..t...... Movies/Action/

Sep 13 12:06:15 Server logger: >f+++++++++ Movies/Action/Saving Private Ryan.mkv

Sep 13 12:07:25 Server kernel: mdcmd (123): clear

Sep 13 12:17:33 Server kernel: mdcmd (124): clear

Sep 13 12:18:27 Server logger: ./Movies/Action

Sep 13 12:18:27 Server logger: .d..t...... Movies/Action/

Sep 13 12:18:27 Server logger: ./Movies/

Sep 13 12:18:27 Server logger: .d..t...... Movies/

Sep 13 12:18:27 Server logger: mover finished

Sep 13 13:06:33 Server kernel: mdcmd (125): spindown 1

Sep 13 13:18:55 Server emhttp: shcmd (81): /usr/sbin/hdparm -y /dev/sdf &> /dev/null

Sep 13 13:19:06 Server kernel: mdcmd (126): spindown 0

Sep 13 13:19:06 Server kernel: mdcmd (127): spindown 4

 

 

Is disk1 in a sign-up group?

dgaschk, I can't find any info on "sign-up group", unless you mean "spin-up group", which I found under disk settings?  Enable spinup groups is set to No.

 

Thanks,

Dave

If his disk was in a spin up group how would that matter? Clearly it would be in a spin of group of its own, otherwise more than one disk would spin up, right?

 

Dave001, I am pretty sure I have seen the same, will test today. I know for sure refreshing the unRAID Web page wakes up random drives (when the web page is left alone fro several hours/days/weeks). I will see if mover does the same for me, sometime today.

 

Is this happening even when the is no data at all to move?

 

or

 

There must be data to move, but its not data that would go to drive 1?

 

OR

 

There must be data to move, and the Share the data is going to also exists on Disk1 as well?

 

Do any of the shares being moved exist on disk 1? The mover reads the directories of all disks included in the share.

Do you have cache_dirs installed? Is disk 1 explicitly excluded from the share that is being moved? If not, it could be the simple act of reading directories that is spinning up disk 1.

Hey BRiT, hows it going, per his comment you probably missed

unRAID Server Pro 5.0-rc6-r8168-test2 - no addons.

 

We'll see what the scenario is when he replies back with a bit more details. As soon as my work on the box ends I will test out the scenario as well.

 

 

 

 

Just tried this, with no data to move ("Mover" did attempted to move what was in my cache Only share that is shared out via SMB/AFP so /mnt/user0 was read, and rsync was triggered (different bug)) and no drive woke up on my system.

 

Dave001, this does sound similar to another report I have filed (previously by myself and others as well)

Topic: Bug - emhttp refresh wakes drives

http://lime-technology.com/forum/index.php?topic=22460.0

 

Could it be that say you have mover enabled and overnight it ran and in the morning you refreshed the web page thus triggering the wake up of the drive or a similar senario?

 

1) You could try this for a quick and easy test, make sure there is nothing to move, then on the main page select/click "spin down" to spin down all the drives, trigger "mover", see if your drive(s) wake?

 

As a second test, place data on the cache drive that you know disk 1 is not apart of that share and repeat the above steps, once "mover" completes you should hopefully see only disk(s) that the data moved to spun up.

 

Let us know.

 

Thanks for the replies. I think I've answered all the questions.

 

Do any of the shares being moved exist on disk 1? The mover reads the directories of all disks included in the share.

 

I only have 4 shares on the system, and each share is on a single disk.

Disk 1 - TV Shows (A-G)

Disk 2 - TV Shows (H-R)

Disk 3 - TV Shows (S-Z)

Disk 4 - Movies

 

 

Could it be that say you have mover enabled and overnight it ran and in the morning you refreshed the web page thus triggering the wake up of the drive or a similar senario?

 

1) You could try this for a quick and easy test, make sure there is nothing to move, then on the main page select/click "spin down" to spin down all the drives, trigger "mover", see if your drive(s) wake?

 

As a second test, place data on the cache drive that you know disk 1 is not apart of that share and repeat the above steps, once "mover" completes you should hopefully see only disk(s) that the data moved to spun up.

 

Let us know.

 

I can refresh the web page with out it spinning up any disks.

I can run the mover when there is no data on the cache drive, and it does not spin up any disks.

I just copied a file to the share "Movies" on disk 4, only the cache drive spun up, then when I run the mover, I had the cache, parity, disk 1 and disk 4 spin up.

 

I'm just on my way out the door at the moment, but when I get home tonight, I'll go back to 5.0-rc5-r8168 and see if the problem follows me back. I never saw it with 5.0-rc5-r8168 previously.

 

Thanks,

Dave

 

 

 

OK, so I will run a very similar scenario, this time I will place data on the cache drive that will need to be moved to the array, and I will pick data that requires to go on disks other than my Disk1 as well, and invoke mover and see if disk 1 wakes up.  Will report back.

 

Just finished the test. You are very correct the mover spins up "DISK 1" when there is actual data to move and none of that data is slated for disk 1. It a bug.

 

My test scenario was just like yours, I have even more drives, 16 data drives in total. I made sure the data to be moved to the array would only hit disk 16 and it woke up disk 1 as well for no reason.

 

The parity drive spinning up is a given.

 

If you have the option and time to test the same with "5.0-rc5-r8168" and post your results that would give Tom a good idea if this was introduced with the latest RC (test2) only or not.

 

Good catch!

 

 

 

 

 

 

Went back to 5.0-rc5-r8168, and the problem followed me, still spinning up disk 1 when I run the mover with data destined for another disk.

 

I don't recall ever seeing this with 5.0-rc5-r8168, prior to upgrading to 5.0-rc6-r8168-test2.

 

 

 

Removed the cache disk from the array, precleared it, added it back to the array and reformatted, just in case there was something left over on the cache disk that was causing it.

Disk 1 still spins up when the mover is run, with data destined for another disk.

 

No idea if it's useful info or not, but there is always 4 reads and 4 writes added to disk 1 when ever the mover is run.

 

are disk1 and disk16 part of the same usershare?

are disk1 and disk16 part of the same usershare?

 

Read the thread, the user share does not include disk 1

Same issue with RC8a

 

Dave if u confirm this as well on RC8a, you can change the thread subject to RC8a. Don't know if Tom is reading them but It would stand out versus leaving it stating 5.0-rc6-r8168-test2

Same issue with RC9a, envoking mover wakes up "DISK1"

 

Simple tes,t stage file(s) that mover will move to a data disk other than DISK1. I went as far as having all clients logged of (SMB and AFP clients). Spin down all drives via web gui and then proceed to envoke mover via WebGui (share settings) you will notice DISK1 is spun up.

 

BTW the web gui never refresh back once mover has completed (mover button stay grayed out), this has been as such for all the RC's I can remember.