Critical error - What should I do?

[BlazeStar]

Don't touch your failing drive - I scrolled all the back to the beginning of the thread and it looks like you don't have any redundancy.

If you try to replace the drive it will kill your pool.

It looks like you have talked about 2 servers (home and office) and I thought you were trying to replace a drive that's part of a mirror (office server).

I won't touch the failing drive, I won't try to replace it.
But thanks for the heads up, you confirmed what I was thinking :)

I bought 5 new 4 TB drives and a LSI Internal SAS SATA 9211-8i card.

So it's all new stuff, not touching anything that's already there.

The server has two "roles"... there is one pool that is used for replicating my office data... this pool is alright... no failing drive or anything.

But the other pool is my personal stuff... and that's the pool that has a drive failing.

It has the system dataset, jails, plugins, shares, etc......

My plan is to :

1. set up a new RAID-Z2 vdev of 5 drives first.
2. Then I want to copy all the files from the existing volume (with the failing drive) to this new volume. I'm thinking to use RSYNC for that. I'm still wondering how to deal with jails and plugins too.
3. Then once everything seems OK, I wanted to destroy the pool with the failing drive.
4. Then take out the failing drive (that's why I needed the "Display drives identification infos" thing.)
5. Then everything is done and my stuff is on the new RAID-Z2 volume... I wanted to re-use the 2 OK drives for something else (I don't know what yet though... because it appears I won't be able to add them to my RAID-Z2 vdev and keep redundancy).

 

[Bidule0hm]

Seems good, but just remember to identify the drives before destroying the pool ;)

You can make a mirror with the 2 good drives.

 

[gpsguy]

You should stop the replication between the office and home.

I believe you could use replication to migrate the data from the existing pool to the new one on the same server. After the old pool is removed (I wouldn't mark them as new) I think you should be able to rename the new pool back to the old one.

These are just high level thoughts. Sorry, I can't help with a detailed game plan.

As you probably know, you'll need to flash your HBA to IT mode. For FreeNAS 9.31 you'll need Phase 20 of the firmware.


Sent from my phone

 

[gpsguy]

Here's a recent thread with details on how to do this - https://forums.freenas.org/index.ph...a-from-one-pool-to-another.40519/#post-255186

I believe you could use replication to migrate the data from the existing pool to the new one on the same server.

 

[BlazeStar]

Seems good, but just remember to identify the drives before destroying the pool ;)

You can make a mirror with the 2 good drives.

A mirror? what do you mean?

Like additional redundancy for my RAID-Z2 vdev of 5 drives?

 

[gpsguy]

I *think* @Bidule0hm was offering up an idea on how you might use the "2 OK drives".

It wouldn't be additional redundancy for your RAIDz2 pool, but rather a 2nd volume (vdev) consisting of both drives in a ZFS mirror. Don't stripe it in with the other pool, leave it as a separate volume. If you do anything that requires scratch (temporary) storage, perhaps you might use that volume instead of your main pool.

I wanted to re-use the 2 OK drives for something else (I don't know what yet though...

A mirror? what do you mean?

 

[Bidule0hm]

Yep, a separate pool with the two drives in mirror ;)

 

[Scharbag]

I thought my drives were running hot (server is in my utility room, it is winter here and the room is quite warm from the furnace and floor heat).

Code:

HDD Temperature:
da0: 35 C - ST4000DM000-1F2168 
da1: 35 C - ST4000DM000-1F2168 
da2: 35 C - ST4000DM000-1F2168 
da3: 34 C - ST4000DM000-1F2168 
da4: 36 C - ST4000DM000-1F2168 
da5: 35 C - ST4000DM000-1F2168 
da6: 38 C - Crucial_CT480M500SSD1 
da7: 38 C - Crucial_CT480M500SSD1 
da8: 35 C - ST4000DM000-1F2168 
da9: 36 C - ST3000DM001-1CH166 
da10: 35 C - ST3000DM001-1ER166 
da11: 36 C - ST3000DM001-1CH166 
da12: 37 C - ST3000DM001-1CH166 
da13: 36 C - ST4000DM000-1F2168 
da14: 36 C - ST3000DM001-1CH166 
da15: 35 C - ST4000DM000-1F2168 
da16: 35 C - ST4000DM000-1F2168 
da17: 36 C - ST3000DM001-1CH166 
da18: 35 C - ST4000DM000-1F2168 
da19: 35 C - ST3000DM001-1ER166 

Definitely get the temps down and as others suggest, get some redundancy. RaidZ2 offers good storage/reliability ratio and is robust. Its performance is more than acceptable for SOHO and media serving.

To preach the gospel though, it never ever ever matters what kind of RaidZn you use if you do not have a backup, you will ALWAYS be at risk of total data loss.

 

[BlazeStar]

Use Bidule0hm's "Display drives identification infos" script to get the serial number. As you've already discovered there's no correlation between the SATA port numbers and adaX.

Just a thought: isn't the output basically the same than if you go in the GUI > Storage and click VIEW DISKS ?

Because I can see the serial numbers there!

Not to complain or anything though, @Bidule0hm your scripts are awesome!


Also, about RAID-Z2: with my 5 X 4 TB drives, and according to this "random" site I found while googling:
http://wintelguy.com/raidcalc.pl

I was expecting 12 Tb of space 60%

Total usable storage capacity (GB) = 12000 *

RAID Type: RAID-Z2 (Double parity with variable stripe width)
Number of RAID groups = 1
Number of drives per RAID group = 5
Total number of drives = 5
Drive capacity (GB) = 4000
Capacity of a single RAID group (GB) = 12000

Space efficiency = 0.6 (60%)
Fault tolerance = 2 disk drives per RAID group

However, I just created my new RAID-Z2 on FreeNAS and it is reporting a little over 10 Tb.

Is that normal?

 

[gpsguy]

The information is in the GUI, but the script presents the information in a useful way.

If zpool status shows an error, you'll need to match the gptid to the serial number.

Just a thought: isn't the output basically the same than if you go in the GUI > Storage and click VIEW DISKS ?

Because I can see the serial numbers there!

In a nutshell, the value you see in FreeNAS is correct. The drive makers use TB, FreeNAS and Windows use TiB. While small, FreeNAS allocates 2GB per disk for swap space. And, FreeNAS has some overhead.

@danb35 filed a bug in an effort to document the calculations. See some of the threads mentioned in this "bug": https://bugs.freenas.org/issues/12968

Also, about RAID-Z2: with my 5 X 4 TB drives, and according to this "random" site I found while googling:
http://wintelguy.com/raidcalc.pl

I was expecting 12 Tb of space 60%

However, I just created my new RAID-Z2 on FreeNAS and it is reporting a little over 10 Tb.

Is that normal?