Upgrading physical disks clarification

[elangley]

Hello All,

Running FreeNAS 9.2.1.3
vol1 consists of 1 vdev with 3 3TB disks (da1, da2, da3) in a RAIDZ1 configuration with 1 50GB SSD drive (da4) for logs

I am planning on upgrading the 3 TB disks to 6TB disks. In reading the documentation (http://doc.freenas.org/9.3/freenas_storage.html?highlight=replace) I see the following steps for my use case:

1a) Backup data
1) In Volume Status set the 1st existing disk to be replaced as "offline"
2) Remove the 1st existing disk
3) Install the 1st upgrade disk
4) In Volume Status click "Replace" on the disk that has been removed and select the newly installed disk
5) Wait for the resilver to complete
6) Repeat until all disks have been replaced.

Am I missing anything?

~eric

 

[BigDave]

8.1.11. Replacing Drives to Grow a ZFS Pool
This section of the manual will be what you need.

 

[DrKK]

Hello All,

Running FreeNAS 9.2.1.3
vol1 consists of 1 vdev with 3 3TB disks (da1, da2, da3) in a RAIDZ1 configuration with 1 50GB SSD drive (da4) for logs

I am planning on upgrading the 3 TB disks to 6TB disks. In reading the documentation (http://doc.freenas.org/9.3/freenas_storage.html?highlight=replace) I see the following steps for my use case:

1a) Backup data
1) In Volume Status set the 1st existing disk to be replaced as "offline"
2) Remove the 1st existing disk
3) Install the 1st upgrade disk
4) In Volume Status click "Replace" on the disk that has been removed and select the newly installed disk
5) Wait for the resilver to complete
6) Repeat until all disks have been replaced.

Am I missing anything?

~eric

Eric:

That's about right. When you finish the third and final drive, your ZFS should autoexpand into the new space.

To recap: offline the disk, fire up the new disk, in the GUI "replace" the old disk with the new disk (should be your only choice for replacement), wait (possibly awhile) for the resilver to complete. Repeat. Autoexpand should happen without your intervention when the process is complete.

 

[gpsguy]

3x6TB disks in RAIDz1 is a dangerous proposition.

You'd be better off backing up the data and creating a new RAIDz2 pool with a minimum of 4x6TB drives.

 

[elangley]

3x6TB disks in RAIDz1 is a dangerous proposition.

You'd be better off backing up the data and creating a new RAIDz2 pool with a minimum of 4x6TB drives.

gpsguy, Thanks I am aware of the risk and I would like to do that. The challenge is that there is remote replication over WAN (backup data) going to this volume that would be have to be started from scratch again (no way) to seed, so I am stuck with what I have for the time being. Only have four drive slots available with one of them being the SLOG.

~eric

 

[elangley]

8.1.11. Replacing Drives to Grow a ZFS Pool
This section of the manual will be what you need.

Thanks, I reviewed that and verified that autoexpand is enabled.

~eric

 

[Robert Trevellyan]

If you have a spare SATA or eSATA port this procedure is easier and safer.

EDIT: never mind, I see you already said you have no spare ports.

 

[elangley]

Robert,

Thanks for this information. It is very helpful. I am going to post it here so if anyone else is looking to the same they won't have to hunt around.

In my case I actually do have a spare SATA port that I can use and just keep swapping disks out as the are "replaced".

Replace procedure per Robert:

If you have a spare SATA port (or an eSATA port and a hard drive dock) you can use the GUI to replace the disks one at a time in a somewhat simpler and safer process than the manual describes for replacing a failed disk:

1. Shut down the system.
2. Install one new disk.
3. Start up the system.
4. Go to the Storage screen, select the pool and click the Volume Status button.
5. In the Volume Status screen, select a disk and click the Replace button.
6. Choose the new disk as the replacement.

When the new disk has resilvered the old one will be automatically offlined. You can then shut down the system and physically remove the replaced disk. One advantage of this approach is that there is no loss of redundancy during the resilver.

 

[Robert Trevellyan]

I've created a feature request topic here to suggest adding this to the documentation.

 

[elangley]

Update: I have installed the first replacement physical disk and used the "Replace" command in Volume/Volume Status. Resilvering is now running and the array is not degraded. Really like that part.

Quick question: Is there anyway to see the progress of the resilver in the shell?

I ran zpool status vol1 but it just shows it resilvering, with no progress report.

~eric

 

[DrKK]

Update: I have installed the first replacement physical disk and used the "Replace" command in Volume/Volume Status. Resilvering is now running and the array is not degraded. Really like that part.

Quick question: Is there anyway to see the progress of the resilver in the shell?

I ran zpool status vol1 but it just shows it resilvering, with no progress report.

~eric

'zpool status' will show you how much of the resilver is yet to go, and approximately how long it will take. Normally the remaining time estimate is an over-estimate, for most cases.

 

[DrKK]

'zpool status' will show you how much of the resilver is yet to go, and approximately how long it will take. Normally the remaining time estimate is an over-estimate, for most cases.

I don't know why you say it's not there. It should be there. Something like "Resilvered 890GB, estimated time remaining 3h2m"

 

[cyberjock]

gpsguy, Thanks I am aware of the risk and I would like to do that. The challenge is that there is remote replication over WAN (backup data) going to this volume that would be have to be started from scratch again (no way) to seed, so I am stuck with what I have for the time being. Only have four drive slots available with one of them being the SLOG.

~eric

So you're saying you are stuck until it blows up in your face, which is possibly going to happen while you are trying to expand the pool.

Well, good luck. :)

 

[elangley]

Okay here's an update.

The reason that I could not see the resilvering stats is because they scrolled off the screen. I ran "zpool status |more" and can now see the status. It would be nice to have this in the front-end.

Following Roberts advice I added one of the new drives to FreeNAS and used the "Replace" function in Volume Status without taking the drive that is being replaced offline first. I had to take another pool offline to allow for this. This allowed the production data pool to remain "Online". If I lost one of the existing drives the pool would become degraded but the data would be intact.

The resilvering of the first drive took 47 hours for 4.90T. Now I am replacing the second drive.

The biggest challenge is the remote replication that is running to the pool, otherwise I would probably just replicate the data to a new pool and then cutover to that pool.

I believe that the current method is as safe as it can be with my environmental limitations. I have a backup for all of the local data. The remote data would not be affected by a local pool failure.

~eric

 

[elangley]

Final notes:

This process is completed and went without a hitch using the "Replace" function three times. Objective safely achieved.

~eric