Pooled drive being ignored, Other data not present but listed.

[Mufti]

Hi all.  Testing out DrivePool and I must say that it's by far the best pooling solution I've ever used.  I am however running into a few oddities.

 

1. A drive in the pool is being completely ignored.

 

- http://imgur.com/O4oODoh

- Drive has been reformatted a few times for troubleshooting, added/removed, etc.  No data present.  Pool folder exists, but is empty.

- It is a different make and model drive than the rest, but it is on the same controller card.

- Functions properly stand alone.

 

 

2. The 'other' data is no where to be found.

 

- I've read every blog post and forum post about the 'other' designation and nothing applies.

- No data outside the pools.

- No obvious hidden data

- Recycling is empty

 

Any assistance would be appreciated.  All settings are default, with added space equalization plugin.

 

Edit:

 

3. No read/write activity is being displayed on the pool anymore.

 

I have reset drivepool by stopping the service and deleting the ProgramData folder.  No change.

[Christopher (Drashna)]

• That's actually completely normal, from the looks of your system.

   

  When new files are put onto the pool, it places them on the disk with the most (absolutely) free space. That means that all of those 3TB drives are going to have to be 900GB filled before DrivePool will start to put data on the 2TB drive.

   

   

• As for the other space:

  http://community.covecube.com/index.php?/topic/37-faq-unduplicated-vs-duplicated-vs-other-vs-unusable/

  5-6GBs sounds about right. This is most likely "Slack space" as defined in the last bullet in my reply in that thread.

  Personally, I have 35TBs and about 10GBs of "other". So that sounds about right.

   

• As for the performance, it doesn't report performance that occurs outside of the pool driver. That means, it only shows performance if you're accessing or writing to the pool.

  If that is the case, and you're not seeing performance, then the "performance counters" on the system have gotten corrupt. To fix that, do this:

  http://wiki.covecube.com/StableBit_DrivePool_Q2150495

   

[Mufti]

In addition to this, the write speed is unusable.  I'm using a dedicated LSI controller for 8 drives, and I can't even break 3MBs write speed.  Every time anything is copied to the pool over the network or locally it crawls.  Direct copies to the drives are fine.  Burst tests was in the 450MBs range.

 

I was really starting to love the product until this stuff. : /

[Christopher (Drashna)]

That is definitely unusual. 

 

Do you have any antivirus or disk tools installed (imaging such as Acronis, or defragging such as PerfectDisk)? If so, they usually install a file system filter, which can cause issues with DrivePool.

Also, do you have "Network IO Boost" enabled? If so, that may be the cause, as it uses additional CPU power to prioritize network IO over local disk IO.

 

Also, have you tried copying file the pool directly (the virtual drive) to another disk? Does that have the same issue? if so, then could you enable file system logging and reproduce the issue (aka, copy files):

http://wiki.covecube.com/StableBit_DrivePool_2.x_Log_Collection

[Mufti]

As an update;  Transfer speed was due to a failing drive in the "array", which was not identified via scanner.  I have since replaced the drive, re-installed the OS and Stablebit Software, and recreated the pool from scratch.

 

Which brings me to another issue; Scanner is not reading S.M.A.R.T. attributes.  They are accessible by every other application I've tried, enabled in the bios, etc.

 

message shows; "The on-disk S.M.A.R.T. check is not accessible on any of your disks.  Drive temperature and mechanical operating parameters are not available and will not be monitored."

 

I've done all the basic trouble shooting, and other drive diagnostic applications can read the s.m.a.r.t. attributes.  The controller has been tested in another system in a different configuration, and I also replaced the controller to see if that was the issue.  It is not likely.

 

Edit; I set 'Smart_NoWmi' to True in the config file (after renaming per the wiki instructions), and now the GUI won't connect to the service.  I've tried saving in both UTF-8 and ASCII.  Same issue.

 

Sigh.

 

Edit 2; Combing through similar threads I came across your Direct I/O tool, and I can retrieve a response using the SCSIPassthrough48 method with Unsafe Direct I/O unchecked.

 

Checking Unsafe Direct I/O reports as well, but... what is Unsafe exactly about this setting?  The wiki describes it but does not go into data corruption, etc.  I assume that since it worked in the diagnostics tool without crashing anything, that it is generally safe to use?

[Christopher (Drashna)]

The "UnsafeDirectIo" setting enables some of the other methods for querying SMART data. The reason that they are marked as "unsafe" is that they are known to cause issues on some controllers. So to make sure that they don't cause issues on a bunch of peoples computers, we mark them as unsafe.

 

If you wish to enable them, then set the "UnsafeDirectIo" option to "True" in your config file. Or use this one:

https://www.dropbox.com/s/0krodbweqz1rrc4/Scanner.Service.exe.config

 

Just be warned, that it can cause issues. Such as crashing the controller (or dropping disks), but as extreme as BSODing. 

ANd as for data corruption, any time a disk drops, it risks corruption. Same with with BSODs. If the system was writing data at the time either of these happen, it can get corrupted as it's getting interrupted. That includes the file allocation table.

 

However, for a lot of controllers, there are no adverse affects to using the settings. I've had a couple of cards that required it to get SMART data.

[4Frame]

Sorry to bump this thread again but I do have a question regarding a pooled drive being ignored.

My main pool consists of 4 Disks with 2x duplication. When I first set up the pool, all data was being balanced equally among the disks. Now after one of the 2TB drives died, I had to remove it from the pool (obviously couldnt do a safe remove since drive was totally dead). 2 days later I added a brand new 2TB drive back into the pool. 

It's still empty and the balancing graph isn't changing. They used to look more "equalized" before.

 

[Christopher (Drashna)]

Well, according to the image, everything is "fine".

 

By fine, I mean that files don't get automatically evened out between the disks. The default file placement strategy is to place files on the disk with the most available free space. It doesn't move existing files unless another balancer (such as Scanner) tells it to move files off of a disk.

 

Though it does look like the feeder disk is slowly balancing data off of it.