I have recently installed 4.3-beta6 on a MD-1500 and am wondering, based on the web interface display, whether I have a problem. 

 

I installed a drive in slot 7, formated it, assigned it as cache and rebooted twice.  From the screen shot attached, you can see that the drive in slot 7 (my cache drive) is red balled and that the overall status of the system is also red.  But the cache function is green and the systems seems to operating OK (with the exception of some skipping on some SageTV HD recordings made after the upgrade). 

 

Is the slot indicator for the drive that is assigned to be cache suppose to be red normally to indicate that that slot has the cache assignment?  If not, is there any clue from my syslog as to what the problem is?

 

Thanks

 

Dudley

Apparently, you assigned your cache drive first to slot 7.  When you did this unRAID made it a part of your array.  When you un-assigned it from slot 7 and used it as the cache drive the array thinks that a drive is missing.  I don't think you should have assigned it to slot 7, or any slot, other than the cache slot.

 

Your array currently thinks a drive has been removed, and in fact, you can still read and write to disk 7.  To get rid of it now you will need to use the "Restore" button (poorly named, since it does not "restore" anything, but instead save a new "initial configuration" based on the currently assigned and working drives, and then immediately throws away any current parity data and set about to re-compute it based on the currently assigned drives.

 

So, stop your array, Check the little checkbox next to the "Restore" button, Press it, and a few hours later your parity will be re-computed (without disk 7) and all the lights will be green, and the slot 7 will simply show as empty.

 

Joe L.

Thanks Joe.  I see the error of my ways now.

 

I stopped the array, check the box under Restore, and hit Restore.  The red ball for slot 7 turned gray, all the other drive indicators turned blue, and the cache drive remained green...all as expected.  However nothing further happened.  No indication of party rebuilding.  It just says "Stopped. Initial configuration".  After waiting awhile, I powered down, rebooted, and repeated but the result was the same. 

Should I pull the cache drive in slot 7 and try to restore again?  I also have the option to "Start" the array but i am a bit hesitant to choose that after reading some earlier posts regarding data getting erased.

 

Dudley

You need to start the array. It will then rebuild parity.

I agree, start the array...

I agree too, start the array...

 

( :P )

 

 

I may be alone in suggesting this, but here goes:

 

Start the array.

 

Remember - I said it first.

Alright so the recommendation to start is unanimous.    Obviously the "Start" command isn't the one that I thought has the potential to erase files.

 

Dudley

The "Format" option, when presented on a drive that shows as "unformatted", but you know has files, is one that will cause data loss.  Unfortunately, whenever unRAID is unable to mount a data drive as a reiserfs file system, it thinks it is unformatted, regardless of the reason.  Clearing and formatting a drive that it thinks is "unformatted" will erase the data on it.  If you see "unformatted" when you think a disk is already formatted, do not check the box to format it.  Capture a syslog, then ask for help before proceeding.  Most times, a simple reboot will get everything back in order.

 

Another way to lose data is to have a data disk fail, and then instead of replacing it and pressing "Start" to have its contents re-constructed, you decide to use the button marked as "restore"  The "restore" button is very poorly named. In reality it saves a new disk configuration to the flash drive based on the currently assigned and working drives.  You do not need to use it when adding a disk to your array, or when replacing a disk.  About the only time it is needed is when you are "removing" a drive from you array and you want to have the array forget it ever existed.  Pressing "restore" will throw away any current parity data and immediately start to calculate parity using only the remaining current working and assigned drives.  Using it when a data drive has failed will eliminate any ability to re-construct the data on that drive from parity, since the old parity data will be gone and the failed drive unreadable too.

 

Of course, a third way to lose data is to have two or more of your data drives fail at the same time.  Only way to try to prevent that is to replace a failed drive as soon as possible, and not wait for a second failure to occur.

 

Joe L.

I wonder since Tom has read SO MANY times that "restore is badly named", if he ever really plans in renaming it.

 

I think if he ever does, we'll refer to the "WR" (with restore) and "AR" (after restore) eras...