Parity Disk Failure

[Ockingshay]

Just came home for lunch and noticed a red ball next to my parity disk. The thing won't spin up and there are many errors in the attached syslog.

 

I have a spare HDD ready to go (although it is currently part of the array, but has zero data on it) but i just wanted to be extra safe and careful with the next set of steps, so that i don't lose anything.

 

Am i correct in doing the following?

 

1) Take the array offline

2) In the drop down menu, select the empty array drive as the new parity drive

3) start up the array and wait for it to rebuild

 

Things i should also be doing?

 

a) I am also thinking i should stop every plugin

b) The last parity check was about 1 month ago

c) should i reboot and load into safe mode?

d) do I need to remove the new parity drive from the array first and then repurpose it? Ie are there any extra steps?

 

z) shall I go and buy a new drive, power off the machine, install the new drive, boot into safe mode and select the new drive as parity?

 

Thanks

i'm running 5.0.5

Parity_Crash.txt

[Ockingshay]

Whatever happens, i needed to take the array offline, so attached is syslog.

 

I've ordered a new drive anyway, just in case. I'll leave the server in this state until i get an experienced reply.

 

Might it not be the drive itself, but the sata port? the parity drive plugs directly into the super micro MB. I have spare ways in my sata card if required.

 

Interestingly, i cannot even select the parity drive. It simply says no device.

Taking_Array_Offline.txt.zip

[itimpi]

The suggested procedure will not work!

 

Once you have added a drive to the array there is no eay way to reduce the number of data drives in the array.    The only relatively easy way is to

• make a note of your urgent assignments;
  
• do a 'new config' to remove all current assignment;
  
• re-assign the drives you want to keep in the array omitting the one you want to now use as parity
  
• start the array and check everything looks like you expect
  
• Stop the array
  
• Assign the parity drive
  
• Start the array rebuilding parity onto the new parity disk.

In theory the last 2 steps could be combined with the previous one, but keeping them separate help protects you from using one of the disks with data you want to keep as the replacement parity drive.

 

The other option is to simply run without parity (I.e. unprotected) until the replacement parity drive arrives and hen follow the steps you mentioned.

[Ockingshay]

Drive will be here tomorrow....I was reading up on the wiki about removing a drive and saw it wouldn't work as expected.

 

I can wait a day to be back online  :'(

 

My only seagate drive as well...about 1 yr old

[itimpi]

My only seagate drive as well...about 1 yr old

I seem to have had an inordinately failure rate with my Seagate 3TB drives - 4 out of 5 starting of fine and passing the pre-clear stress test but then developing a significant number of re-allocate sectors after a month or two of use in the array.  My WD Green 3TB drives on the other hand have had 3 out of 5 DOA (or failing during initial pre-clear), but then the drives that were OK and those from the RMA operating flawlessly (so far at least).  Given a choice I much prefer the WD type of failure as it happens before I have committed live data to them.

[Ockingshay]

All my other drives are WD red, as is this new one on the way.

 

I wanted to double check a cable hadn't come loose, so after physically going over the server and rebooting into safe mode there is a blue ball next to the existing parity drive and it says that it is a new parity drive and needs rebuilding.

 

I just really hope this comes back online as it was...how long does a rebuild take? there is about 20TB of data on there across 10 drives.

 

I do have backups just in case, bloody thing.... 

[itimpi]

All my other drives are WD red, as is this new one on the way.

 

I wanted to double check a cable hadn't come loose, so after physically going over the server and rebooting into safe mode there is a blue ball next to the existing parity drive and it says that it is a new parity drive and needs rebuilding.

 

I just really hope this comes back online as it was...how long does a rebuild take? there is about 20TB of data on there across 10 drives.

 

I do have backups just in case, bloody thing.... 

Rebuilding parity depends primarily on the size of the parity drive as every sector on it needs re-writing.    It has nothing to do with how much data you have on the drives as it is a simple sector level activity that ignores any filing system on the disks.