5.0b4: LSI1068E - Drive spindown/spinup


Recommended Posts

I'm not sure if you're aware of this or not, but there's a slight issue with spindown and spinup within emhttp/webGUI for drives on the LSI 1068E-based controllers.

 

As far as I can tell, hdparm -C and hdparm -Y and hdparm -y function fine.

 

Feb  9 16:43:37 reaver kernel: mdcmd (2654): spinup 0

Feb  9 16:43:37 reaver kernel: mdcmd (2655): spinup 1

Feb  9 16:43:37 reaver kernel: mdcmd (2656): spinup 2

Feb  9 16:43:37 reaver kernel: mdcmd (2657): spinup 3

Feb  9 16:43:37 reaver kernel: mdcmd (2658): spinup 4

Feb  9 16:43:37 reaver kernel: mdcmd (2659): spinup 5

Feb  9 16:43:37 reaver kernel: md: disk4: ATA_OP_SETIDLE1 ioctl error: -22

Feb  9 16:43:37 reaver kernel: md: disk5: ATA_OP_SETIDLE1 ioctl error: -22

 

Feb  9 16:45:11 reaver kernel: mdcmd (2660): spindown 0

Feb  9 16:45:12 reaver kernel: mdcmd (2661): spindown 1

Feb  9 16:45:13 reaver kernel: mdcmd (2662): spindown 2

Feb  9 16:45:14 reaver kernel: mdcmd (2663): spindown 3

Feb  9 16:45:15 reaver kernel: mdcmd (2664): spindown 4

Feb  9 16:45:15 reaver kernel: md: disk4: ATA_OP_STANDBYNOW1 ioctl error: -22

Feb  9 16:45:15 reaver kernel: mdcmd (2665): spindown 5

Feb  9 16:45:15 reaver kernel: md: disk5: ATA_OP_STANDBYNOW1 ioctl error: -22

Feb  9 16:45:15 reaver kernel: mdcmd (2666): spindown 4

Feb  9 16:45:15 reaver kernel: md: disk4: ATA_OP_STANDBYNOW1 ioctl error: -22

Feb  9 16:45:17 reaver kernel: mdcmd (2667): spindown 4

Feb  9 16:45:17 reaver kernel: md: disk4: ATA_OP_STANDBYNOW1 ioctl error: -22

 

^ The last 2 line sets keep cycling as it tries to repeatedly spin down the drive ^

 

hdparm -C /dev/sdg

 

/dev/sdg:

drive state is:  active/idle

root@reaver:/cache/.downloads/extracted# hdparm -Y /dev/sdg

 

/dev/sdg:

issuing sleep command

 

# A clicking noise like the hard drive parking is heard

 

root@reaver:/cache/.downloads/extracted# hdparm -C /dev/sdg

 

/dev/sdg:

drive state is:  standby

 

root@reaver:/cache/.downloads/extracted# smartctl -A /dev/sdg

 

# A clicking noise and a motor spinning noise is heard

 

root@reaver:/cache/.downloads/extracted# hdparm -C /dev/sdg

 

/dev/sdg:

drive state is:  active/idle

 

hdparm -y /dev/sdg

 

/dev/sdg:

issuing standby command

 

# a slight clicking noise is heard

 

root@reaver:/cache/.downloads/extracted# hdparm -C /dev/sdg

 

/dev/sdg:

drive state is:  standby

 

root@reaver:/cache/.downloads/extracted# smartctl -A /dev/sdg

 

# A clicking noise and a motor spinning noise is heard

 

root@reaver:/cache/.downloads/extracted# hdparm -C /dev/sdg

 

/dev/sdg:

drive state is:  active/idle

Link to comment

This is a known issue being mentioned in the announcement thread.

 

The problem is the ATA command that the unraid driver is using to spin drives up/down.  Apparently the driver for that card does not do any translation of ATA commands to attached SATA or SAS drives (unlike mvsas driver for example).  This is the reason scsi has not been supported in the past because of these kinds of issues...

 

Why does unraid driver spin drives up/down?  In order to (efficiently) implement spinup groups.

 

So I'm thinking of redesigning the entire spinup/spindown/spinup groups feature, to solve this problem, as well as implement a feature request of tying spinup groups to user shares.  So here's what I'm contemplating:

 

a) Tear out all current spinup/spindown from driver and management code.

b) Program the spin-down delay in the drives themselves - if you are using a (really) old drive that doesn't support internal spin down timer - time to replace that old drive anyway.

c) Implement spinup groups only in the user share file system code.  This means if you are using spinup groups to spinup both master/slave disks on same IDE channel via access to a disk share - this won't work anymore.

 

Doing all this makes the code quite a bit simpler, adds some functionality, fixed scsi spin issues, but removes functionality detailed in c) above.

 

So... anyone using spin up groups for spinning up both disks of an IDE channel???

Link to comment

I guess this thread is then more for those who missed it in the announcement thread :-[.

 

I can understand how there's several intricacies in trying to expand features on long lived software especially in the area where the underlying software and hardware technology can change quite drastically.

 

That's some interesting possibilities there for spinup changes. Did you want to continue discussion in this thread or rather refocus it into one under the Roadmap forum to garner feedback from others who might not see this thread due to the title and nature of it?

Link to comment

So... anyone using spin up groups for spinning up both disks of an IDE channel???

Yes, those of us with a mostly IDE based server that we've had been using for about 5+ years now.  ;)

 

However, perhaps it could be implemented at the user share level regardless.

 

If I had a user-share for my "music" I could put in its spin-up group all the disks with a top level folder of "music" AND the other disks on the appropriate IDE channels they may share.

 

So, music directories might exist on disk1, disk2, and disk3.  Disk1 and 2 might share an IDE channel.  Disk3 might share one with disk4.  My spin-up group list for "music" would then be disk1,disk2,disk3,disk4.

 

Those who would complain are those who are accessing disk shares directly.

 

Joe L.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.