i am hoping someone can help me since the first rule of unraid is not to do something stupid in a panic.

 

i lost one of my hard drives due to a drive failure and pulled out the defective drive and replaced it with a new drive (while the array was active, which was a stupid move in hindsight because i started getting reiserfs and i/o errors on the syslog afterwards).

 

anyways, on the advice of some smart people like dgaschk here, i stopped the array, unselected the drive that was defective from the array panel, and left the array stopped until the new drive had completed preclear.

 

now here's the problem. somehow in the last few hours, i lost access to the webgui. emhttp is running and i tried to kill the process and restart it but nothing happened. i continue to get intermittent i/o and reiserfs errors, but all the drives seem to be fine.

 

i'm hopeful that some of this will be corrected when i reboot the system. but what i worry about is that since i will have to do a hard reboot, whether unraid will restart and try to rewrite parity based on the incomplete array since there is a drive that is unassigned now.

 

further, what happens if i restart and don't have emhttp? then parity could be rebuilding in the background and i would have no way of stopping it!

 

can someone tell me is there a way to check if parity is rebuilding and kill the process if i don't have access to the webgui on my hard reboot?

 

Ouch!  Hot-swapping drives in UnRAID is a no-no !!

 

Not sure what state you're in;  and, like you, I'm not certain what would happen if you booted now -- I know UnRAID will want to do an automatic parity check, but don't know what the absence of a drive would do to that.

 

Here's what I'd do if I had this situation:

 

(a)  Disconnect ALL drives

(b)  Boot to the UnRAID flash drive; go to the Web GUI; and set a New Config on the Utils tab with NO drives assigned.

©  Shut down

(d)  Re-connect ALL drives, including the bad one

(e)  Boot to the UnRAID flash drive; go to the Web GUI; and set a New Config with ALL of the drives assigned (including the bad one but NOT the new one you want to replace it with) AND check the "Trust Parity" option.  Start the array - if it starts a parity check IMMEDIATELY stop it -- and then Stop it.

(f)  Unassign the bad drive; then Start the array so it shows a missing drive.

(g)  Shut down and replace the bad drive with the new one.  [You can skip this step if the new drive is on it's own SATA port and is already connected]

(h)  Boot to UnRAID, Stop the array if it auto-starts; then assign the new drive in place of the missing one.

(I)  Start the array and let it rebuild your disk.

 

When it finishes the rebuild, do a non-correcting parity check to confirm all went well.

 

 

Garycase, back to my rescue.. Thank you!

 

I have two questions about your answer.

 

Step (e) -- when you say "IMMEDIATELY Stop It... Then Stop It" -- I assume that the first "it" refers to stopping the parity check, and the second it is stopping the array, correct? It might be obvious, but I don't want to leave anything to miscommunication.

 

final Step -- How do I do a "non-correcting parity check"? I don't have access to my webgui so I can't remember if there is an option to do that.

 

Thank you again so much! I will let pre clear finish its work with the drive and retry again.

 

I tried stopping the array and re-invoking emhttp but I get the following:

 

Segmentation Fault [core dumped]

So I went back and stopped the array from the command prompt, and illustrated by Prostuff1 in a prior post here using:

 

/etc/rc.d/rc.unRAID stop

Since then I haven't seen any errors in my syslog and all has been quiet. After my pre clear is done I will try your strategy and see what comes up.

Thanks again, Garycase.

 

 

 

 

Here's what I'd do if I had this situation:

 

(a)  Disconnect ALL drives

(b)  Boot to the UnRAID flash drive; go to the Web GUI; and set a New Config on the Utils tab with NO drives assigned.

©  Shut down

(d)  Re-connect ALL drives, including the bad one

(e)  Boot to the UnRAID flash drive; go to the Web GUI; and set a New Config with ALL of the drives assigned (including the bad one but NOT the new one you want to replace it with) AND check the "Trust Parity" option.  Start the array - if it starts a parity check IMMEDIATELY stop it -- and then Stop it.

(f)  Unassign the bad drive; then Start the array so it shows a missing drive.

(g)  Shut down and replace the bad drive with the new one.  [You can skip this step if the new drive is on it's own SATA port and is already connected]

(h)  Boot to UnRAID, Stop the array if it auto-starts; then assign the new drive in place of the missing one.

(I)  Start the array and let it rebuild your disk.

 

When it finishes the rebuild, do a non-correcting parity check to confirm all went well.

That was poor sentence construction on my part in (e) 

 

But you correctly figured out what I meant.    I should have said something like:  Start the array (so it "sees" the New Config) and then Stop it.    If it happens to start a parity check when you start the array, IMMEDIATELY abort that parity check.

 

 

r.e. non-correcting checks => As you may know, I am generally NOT a fan of doing non-correcting checks ... but after a disk rebuild is the exception.  At that point you DO want to use the non-correcting option, so if the rebuild didn't go well, you can do it again.    On the Web GUI, just under the box you click on to start a parity check, there's a box that's normally checked that says "Correct any Parity-Check errors by writing the Parity disk with corrected parity."    To do a non-correcting check, you UNCHECK that box before starting the check.