Drive SMART Fail - Replace Now or Switch Version First?

[indivision]

I have a degraded volume with what appears to be a broken drive. SMART fails for that drive now.

I have a replacement drive ready to go. But, I'm wondering if I should do that replacement now (I'm running Corral) or would it be safe to switch to the latest 11 build first, then replace the drive?

To take a few steps back, is there anything that can be done to double-check the health of that drive? Or, if it is failing SMART, is it pretty safe to say that it's toast?

 

[Artion]

Maybe this could be helpful. Try to identify the exact SMART error by executing smartctl and then follow some tips from that thread. IMHO the sooner you change the drive the better. If there are chances for trouble switching from FN Corral to FN11, there will be more with a degraded pool.

 

[indivision]

Maybe this could be helpful. Try to identify the exact SMART error by executing smartctl and then follow some tips from that thread. IMHO the sooner you change the drive the better. If there are chances for trouble switching from FN Corral to FN11, there will be more with a degraded pool.

Thank you. Gives me a direction.

It looks like the bad drive is removed. So, not available at "ada3" like in your link. Where could I find the reference to this drive to run smartctl on?

Do you have a link to any directions covering how to replace the drive via CLI? Last time I did this I was using a much older version of FreeNAS and used GUI. Not sure how it is done with Corral...

 

[Artion]

If its not available maybe it is gone! You can try reboot your machine and see if it gets available again. If not then you can use the zfs or zpool command in the shell to replace it with a new one, if the option isn't in the GUI. (I remember there are also commands in CLI that do the same thing) Sorry but I have no box with Corral anymore. You may find something here or maybe there is an online documentation in your install.

 

[melloa]

I'm running Corral

Have drive replacement in Corral be fixed? I remember people having problems with that around March/2017, so not too hopeful has be fixed. If you have access to your data, please back-up and move to another build, try to import and replace. If won't work you can always start from scratch with the back-up. I'm on 9.10 waiting for the dust to settle on 11 to upgrade my production NAS(es).

I think @Ericloewe wrote something about moving from Corral to 9.10 but I can't find it. Maybe some one can help?

 

[Ericloewe]

Have drive replacement in Corral be fixed? I remember people having problems with that around March/2017, so not too hopeful has be fixed. If you have access to your data, please back-up and move to another build, try to import and replace. If won't work you can always start from scratch with the back-up. I'm on 9.10 waiting for the dust to settle on 11 to upgrade my production NAS(es).

I think @Ericloewe wrote something about moving from Corral to 9.10 but I can't find it. Maybe some one can help?

Resources section.

https://forums.freenas.org/index.ph...ng-from-freenas-corral-to-freenas-9-10-11.36/

 

[indivision]

If its not available maybe it is gone! You can try reboot your machine and see if it gets available again. If not then you can use the zfs or zpool command in the shell to replace it with a new one, if the option isn't in the GUI. (I remember there are also commands in CLI that do the same thing) Sorry but I have no box with Corral anymore. You may find something here or maybe there is an online documentation in your install.

Aha. Here is the how-to: https://forums.freenas.org/index.php?threads/how-to-storage-replace.53925/

Thank you. FWIW, the GUI shows the drive as under folder "gptid/" but it cuts off the full path. Within that folder there were what looked like some kind of drive references. But, smartctl -a didn't work on them.

Have drive replacement in Corral be fixed? I remember people having problems with that around March/2017, so not too hopeful has be fixed. If you have access to your data, please back-up and move to another build, try to import and replace. If won't work you can always start from scratch with the back-up. I'm on 9.10 waiting for the dust to settle on 11 to upgrade my production NAS(es).

I think @Ericloewe wrote something about moving from Corral to 9.10 but I can't find it. Maybe some one can help?

Thank you. From what I read, the issue wasn't that drive replacement didn't work. Just that there wasn't a GUI for it.

Moving to another build should be easy. So, you advise doing that with a degraded pool? Then replacing after? (I already have a back-up of the data. But, it would be less fun recovering from that than replacing the drive. :) ).

 

[melloa]

Aha. Here is the how-to: https://forums.freenas.org/index.php?threads/how-to-storage-replace.53925/

Thank you. FWIW, the GUI shows the drive as under folder "gptid/" but it cuts off the full path. Within that folder there were what looked like some kind of drive references. But, smartctl -a didn't work on them.



Thank you. From what I read, the issue wasn't that drive replacement didn't work. Just that there wasn't a GUI for it.

Moving to another build should be easy. So, you advise doing that with a degraded pool? Then replacing after? (I already have a back-up of the data. But, it would be less fun recovering from that than replacing the drive. :) ).

I just simulated a volume fail and it didn't work from the cli for me.

The command should be:

volume <your_volume_name> replace <old_hdd> <new_hdd>

In my case I'm getting the error "NoneType object is not subscriptable".

Not sure if caused 'cause I'm testing on a VM, but if you want to try, check: https://www.youtube.com/watch?v=yK-SFLGottk

 

[indivision]

Ok. I decided to install FreeNAS 11 and look into replacing the drive there. But, it seems to show even less information than Corral did.

The bad drive does not appear available. And when I try to "import volume" the volume I need is not listed.

Any advice on how to proceed from here? Any guides on how to do this replacement via console?

 

[danb35]

And when I try to "import volume" the volume I need is not listed.

Is the pool already imported? If so, it won't be available to import again.

The bad drive does not appear available.

If the drive is completely dead (which is what it sounds like), it won't show in the View Disks page.

What's the output of zpool status and zpool import? In code tags, please.

 

[indivision]

Is the pool already imported? If so, it won't be available to import again.

If the drive is completely dead (which is what it sounds like), it won't show in the View Disks page.

What's the output of zpool status and zpool import? In code tags, please.

Thank you for taking a look.

The pool is not already imported, unfortunately.

I agree about the drive. I think that it's completely dead. Doesn't show on the View Disks page (even though the other drives do).

zpool status:

Code:

pool: freenas-boot																											   
state: ONLINE																													 
  scan: none requested																											 
config:																															 
																																	
		NAME		STATE	 READ WRITE CKSUM																					 
		freenas-boot  ONLINE	   0	 0	 0																					
		  mirror-0  ONLINE	   0	 0	 0																					 
			da0p2   ONLINE	   0	 0	 0																					 
			da1p2   ONLINE	   0	 0	 0																					 
																																	
errors: No known data errors

zpool import:

Code:

pool: optimus																													
	 id: 3732643262260824293																										
  state: DEGRADED																												   
status: The pool was last accessed by another system.																			 
action: The pool can be imported despite missing or damaged devices.  The														 
		fault tolerance of the pool may be compromised if imported.																 
   see: http://illumos.org/msg/ZFS-8000-EY																						 
config:																															
																																	
		optimus										 DEGRADED																	
		  mirror-0									  ONLINE																	 
			gptid/ef276eb5-e507-11e3-a0e9-3085a93e10d2  ONLINE																	 
			gptid/f05b2144-e507-11e3-a0e9-3085a93e10d2  ONLINE																	 
		  mirror-1									  DEGRADED																	
			gptid/09840d8d-e508-11e3-a0e9-3085a93e10d2  ONLINE																	 
			1349034441408848568						 UNAVAIL  cannot open   

 

[danb35]

And optimus doesn't show up when you go to Import Volume? Try zpool import -f optimus followed by zpool export optimus, and try importing through the GUI again.

 

[indivision]

It doesn't show up when I go to import volume. I read in another thread that this happened to someone else on 11. They went back to 9.10.4 and it worked...

Is the "-f" option safe? I see these warnings about using it and want to be careful.

 

[danb35]

The -f flag is safe, the -F flag is potentially destructive. The -f flag just lets you import the pool on a different system than the one that last used it.

 

[indivision]

Ok. I ran the command with -f, export, then tried to import with GUI. No change. The list of pools to import is just blank.

 

[danb35]

Do you have a spare USB stick that you can try to install 9.10.2 on? If so, try installing that and see if your pool will show in the Import Volume window--but if it does, don't proceed with the import. Boot back into your FN11 install, and if the pool still doesn't show, go to the Support button in the web GUI and file a bug report--this will attach your debug information to the bug, so the devs will hopefully have a better chance of fixing it, or at least figuring out what's going on.

 

[indivision]

Ok. I don't think I want to mess with 11 if it has bugs at that level. If 9.10.2 works, I will probably just stick with that for a while.

 

[danb35]

I don't think I want to mess with 11 if it has bugs at that level.

Fair enough, though I doubt that importing degraded pools is something that would have been deliberately tested. If the pool shows with 9.10, the bug report on 11 would still be helpful, even if you're then going to go back to 9.10.2.