I had 9 data disks - one of which (the smallest) was a PATA drive. My system has been stable for a looong time, but I recently ran out of space so procured a 3Tb SATA as a replacement. Before I started the process there were no errors and parity was good.

 

The only free slots were on a PCI-x controller. The first port I connected it to made boot take ages, and a number of IO errors on the screen. Shut it down and moved it to the other port and all looked good, so proceeded with the standard upgrade process.

 

Web interface and shares were unresponsive for a period, however they came back and the process seemed to progress well. A refresh this morning shows it completed, however 2x drives (one of which is on the same controller) are now reporting massive error counts.

 

Shares are up, however some won't open and those that do have almost no data in them. UPDATE: At the console I see a screen full of REISERFS errors for "(device md1)"

 

I'm guessing that the problem is the PCI-x card (possibly in combination with mobo IRQ?) and that best course may be to rollback. I am nervous because parity seems to have been written to - is it as simple as pulling the new drive and replacing the old?

 

I am still using 5.0 RC12a. Screenshot attached. Syslog is at 78Mb and still generating so I haven't attached yet. Will do so when it completes if anyone thinks it might be useful.

 

And now the web interface (and unmenu) is not responding.

 

Should I follow this guide and remove the new drive?: http://lime-technology.com/wiki/index.php/UnRAID_Manual_5#Remove_one_or_more_data_disks

 

Then re-add the old drive and create a new config?

The New Configuration utility resets the array configuration.

Thankyou - I have swapped the drives back, initiated 'new config' and put the drives back in the same order. Parity rebuild is running, and my data seems to be back.

 

While I am here, is it worth upgrading unraid?

 

Update to 5.0.5. The beta versions all have serious bugs.

Thankyou, I'll do that ASAP.

 

I also want to remove the PATA drive (don't want to be caught in a similar position in the future should it fail). I will then pull one of the 1Tb SATA drives and replace with the new 3Tb drive.

 

I did a search and found 2 very different approaches:

[*]This one discusses removing from shares, renaming folders, and manually moving data: http://lime-technology.com/wiki/index.php/Shrink_array

[*]This one moves the data to the cache drive, and creates a new config with the (old data) drive removed from the array: http://lime-technology.com/wiki/index.php/FAQ_remove_drive

 

Both also use MC instead of windows explorer to move the data - is this just for performance (so that the data transfer is local to the unraid server, and not across the network)?

 

The second option seems the simplest. Is there a preferred (or even alternative) method?

Either method looks ok. You can export disk shares to drag and drop the files on a client PC. As long the PC and network are stable this should not take any longer than making copies local on the server.