Hard Drvies Device Name

[guillaumesoucy]

Hi everyone,

I have a question who may sound stupid but I will ask it, With Unix and Linux devices like hard drives are named by example: sda, sdb, and there come the partitions: sda1, sda2, sdb1 sdb2 but in the case I remove an hard disk because the simple reason, he his cashed, and this drive has the device name sdb, when the new disk mounted in FreeNAS will this drive will take the device name sdb as the old drive does? If yes what happen if two drives are replaced, sdb and sde but no replaced in the same order?

Thanks,

Guillaume

 

[Bidule0hm]

Device labels are not linked to the physical device, they are just incremental (like you plug a drive then his label will be xxa or xx0, you plug another then his label will be xxb or xx1, etc...).

That's why FreeNAS only cares about the GPTID (written to the physical device by the system) and you only want to link GPTID with serial number to identify a drive as the device label can change from reboot to reboot.

 

[depasseg]

The simple answer is to think of those names as completely random and then don't do anything that depends on those names.

Sent from my Nexus 9 using Tapatalk

 

[guillaumesoucy]

Hi guys!

Its make sense. For real its a smart way to do the things. So, if I replace the drive in bay 2 and the drive in bay 4 for example, what will tell me the which drive is which one when I recreate the mount points?

 

[gpsguy]

If I recall correctly, your pool(s)/layout(s) are much different than most of the rest of us.

Please post the results of zpool status in code tags (to preserve formatting), so we can see what you have.

 

[guillaumesoucy]

Code:

  pool: freenas-boot
state: ONLINE
  scan: scrub repaired 0 in 0h1m with 0 errors on Sat Mar  5 03:46:10 2016
config:

    NAME                                          STATE     READ WRITE CKSUM
    freenas-boot                                  ONLINE       0     0     0
      gptid/984a078e-012a-11e5-8f49-00219b11fd1f  ONLINE       0     0     0

errors: No known data errors

  pool: zawack1
state: ONLINE
  scan: scrub repaired 0 in 3h43m with 0 errors on Sun Mar 20 03:43:37 2016
config:

    NAME                                          STATE     READ WRITE CKSUM
    zawack1                                       ONLINE       0     0     0
      gptid/31fff82a-0384-11e5-abbc-b8975a80a7c7  ONLINE       0     0     0

errors: No known data errors

Et voilà! The server can contain 24 disks but for the moment it have only one.

 

[gpsguy]

I still don't understand your rationale for creating 24 independent pools consisting of a single disk in each one. There's no fault tolerance.

@Bidule0hm provided the answers above.

If one of your pools fail, you need to map the gptid back to the serial number of the drive. He's got a script that might be helpful for you. See: https://forums.freenas.org/index.ph...d-identification-and-backup-the-config.27365/ and review the "Display drives identification infos"

 

[depasseg]

Hi guys!

Its make sense. For real its a smart way to do the things. So, if I replace the drive in bay 2 and the drive in bay 4 for example, what will tell me the which drive is which one when I recreate the mount points?

Storage -> View Disks, which will tell you the daXX to serial number. Alternatively, you can use dd to read a disk and look for the activity light.

 

[guillaumesoucy]

@depasseg

I saw when I delete a pool (1 drive) and I re-add it after I saw also a long serie of numbers when its time to select a device. But I have hard time to figure out which drive it is when news drives are added. It can be simple but except for me.

I will think to this for trying to figure and get back with any questions.

Have a great day.

 

[pirateghost]

@depasseg

I saw when I delete a pool (1 drive) and I re-add it after I saw also a long serie of numbers when its time to select a device. But I have hard time to figure out which drive it is when news drives are added. It can be simple but except for me.

I will think to this for trying to figure and get back with any questions.

Have a great day.

Your decision of using one disk per pool is a waste of FreeNAS' talent...

 

[gpsguy]

As pirateghost implied it's difficult for you, due to the 1 disk per pool. I doubt we have any other forum member who have a 24 disk chassis with 24 volumes.

When you detach the volumes do you mark them as new? Or are you retaining the data?

Please tell us what you are doing and/or trying to accomplish.

For example, do you have some sort of service company and have 40 customers? For each customer you buy a 6TB drive. With 24 bays in the server you have access to 24 clients data. If you need data from one of the other 16 clients, you need to detach a volume and import one of the ones off the shelf.

I made up the example above.

Please explain what you are doing.


Sent from my iPhone using Tapatalk

 

[guillaumesoucy]

@gpsguy

The only time I remove a drive is when he is crashed, otherwise drives stay in the chassis until he die. When I replace a drive, he is marked as new since its a brand new drive. The only inquiry I get with this is, if I replace multiples drives like two drives, one in bay #2 and another in bay 4# by example. Each drive is mounted in "/mnt" so drive 1 is mounted as "zawack1" (as showed with the command zpool status), drive 2, not present at the moment but will be mounted as "zawack2". When I insert the new drive in bay #4 who will told me which drive is it in the device list for I can mount it as "zawack4" in "/mnt" like the old one who was in bay #4 before he die? For avoiding of mounting it as "zawack2" because "zawack2" is the drive who will be replaced in bay #2.


@pirateghost

For the redundancy, I do backup copy 1:1 each day at the evening and I store it in a heated and secured locker near my town when I go to work at midnight, its on my way so its not a matter for me. I don't trust RAID, but if I can get another FreeNAS server with the same amount of disk space, I likely to do some kind of RAID at my place and doing the same level of RAID at the remote location and doing a replication task. I can't do this, unfortunately its not possible.

 

[pirateghost]

LMAO @ not trusting raid....it's used by every major storage and server platform in the world.

Having a pool per disk is just stupid. But whatever. You never truly listen to what others say anyway.

 

[guillaumesoucy]

I work in that way since ever and I never get issues.

 

[pirateghost]

I work in that way since ever and I never get issues.

Sure.

 

[depasseg]

I work in that way since ever and I never get issues.

Except that you need to manually correlate slots to disk names to disk GUID's whenever you need to replace one.

 

[guillaumesoucy]

Yes right but I just find a simple way to do the things. Removing and replacing one drive at time can do the task easier and it not taking more time to accomplish. Simple as that.

 

[depasseg]

I think the 16 previous posts beg to differ with calling it "easier".

 

[pirateghost]

Yes right but I just find a simple way to do the things. Removing and replacing one drive at time can do the task easier and it not taking more time to accomplish. Simple as that.

You think that recreating a pool and the necessary datasets is EASIER than letting a redundant disk handle it for you? You think it takes LESS time?

I don't know what to think about asinine statements like that but I am going to guess you've never lost a pool on freenas before.