SOLVED Copying One Pool to Another

[Chris Tobey]

Hello,

I am having an issue where when I do a zfs send and receive, the receive side accepts all the data, creates the datasets, shows the space as consumed, but I cannot see any files.

Here is what I do:

My two pools use the same drives, and have the same capacity. One is a RAIDZ-1 and one is a RAIDZ-2.
Pool names are SG1 (RAIDZ-1) and SG2 (RAIDZ-2).

~#zfs snapshot -r SG1@backup
~#zfs send -Rv SG1@backup | zfs receive -Fdu SG2

This command appeared to work successfully and copied my 10.8T in about 24 hours.

SG2 is mounted as it should be and appears as /mnt/SG2, but when I navigate to it there are no files. The attached image shows what I see in the FreeNAS GUI.

If anyone could shed some light on this it would be greatly appreciated.

For reference, I am on FreeNAS 9.2.1.4.
SG1 pool was originally created on FreeNAS 9.1.1 (I think)
SG2 pool is new and created with FreeNAS 9.2.1.4.

Edit:

SOLUTION: Detach the pool and auto-import it.

 

[cyberjock]

I just did this less than 48 hours ago with 9.2.1.9 and it worked just fine.

Any chance you could try a reboot?

Also, if you look at that screenshot it doesn't look like its transferred data. The used columns are incorrect. ;)

 

[Chris Tobey]

Hi cyberjock,

I agree, it doesn't look like it transferred anything. I know it did, and my pools are 16TB, so obviously some data is used.

Maybe a bug in 9.2.1.4?

Here are my zfs get alls:

Code:

[root@stargate] /mnt/SG2/users# zfs get all SG1
NAME  PROPERTY  VALUE  SOURCE
SG1  type  filesystem  -
SG1  creation  Mon Nov 25 14:48 2013  -
SG1  used  6.58T  -
SG1  available  7.67T  -
SG1  referenced  435K  -
SG1  compressratio  1.76x  -
SG1  mounted  yes  -
SG1  quota  none  default
SG1  reservation  none  default
SG1  recordsize  128K  default
SG1  mountpoint  /mnt/SG1  default
SG1  sharenfs  off  default
SG1  checksum  on  default
SG1  compression  off  default
SG1  atime  on  default
SG1  devices  on  default
SG1  exec  on  default
SG1  setuid  on  default
SG1  readonly  off  default
SG1  jailed  off  default
SG1  snapdir  hidden  default
SG1  aclmode  passthrough  local
SG1  aclinherit  passthrough  local
SG1  canmount  on  default
SG1  xattr  off  temporary
SG1  copies  1  default
SG1  version  5  -
SG1  utf8only  off  -
SG1  normalization  none  -
SG1  casesensitivity  sensitive  -
SG1  vscan  off  default
SG1  nbmand  off  default
SG1  sharesmb  off  default
SG1  refquota  none  default
SG1  refreservation  none  default
SG1  primarycache  all  default
SG1  secondarycache  all  default
SG1  usedbysnapshots  409K  -
SG1  usedbydataset  435K  -
SG1  usedbychildren  6.58T  -
SG1  usedbyrefreservation  0  -
SG1  logbias  latency  default
SG1  dedup  off  default
SG1  mlslabel  -
SG1  sync  standard  default
SG1  refcompressratio  1.00x  -
SG1  written  0  -
SG1  logicalused  10.9T  -
SG1  logicalreferenced  50K  -
[root@stargate] /mnt/SG2/users# zfs get all SG2
NAME  PROPERTY  VALUE  SOURCE
SG2  type  filesystem  -
SG2  creation  Sun Jan 11 22:41 2015  -
SG2  used  6.65T  -
SG2  available  7.61T  -
SG2  referenced  543K  -
SG2  compressratio  1.75x  -
SG2  mounted  yes  -
SG2  quota  none  default
SG2  reservation  none  default
SG2  recordsize  128K  default
SG2  mountpoint  /mnt/SG2  default
SG2  sharenfs  off  default
SG2  checksum  on  default
SG2  compression  lz4  local
SG2  atime  on  default
SG2  devices  on  default
SG2  exec  on  default
SG2  setuid  on  default
SG2  readonly  off  default
SG2  jailed  off  default
SG2  snapdir  hidden  default
SG2  aclmode  passthrough  local
SG2  aclinherit  passthrough  local
SG2  canmount  on  default
SG2  xattr  off  temporary
SG2  copies  1  default
SG2  version  5  -
SG2  utf8only  off  -
SG2  normalization  none  -
SG2  casesensitivity  sensitive  -
SG2  vscan  off  default
SG2  nbmand  off  default
SG2  sharesmb  off  default
SG2  refquota  none  default
SG2  refreservation  none  default
SG2  primarycache  all  default
SG2  secondarycache  all  default
SG2  usedbysnapshots  1.55M  -
SG2  usedbydataset  543K  -
SG2  usedbychildren  6.65T  -
SG2  usedbyrefreservation  0  -
SG2  logbias  latency  default
SG2  dedup  off  default
SG2  mlslabel  -
SG2  sync  standard  default
SG2  refcompressratio  1.00x  -
SG2  written  160K  -
SG2  logicalused  10.8T  -
SG2  logicalreferenced  56K  -

 

[Chris Tobey]

Sadly I cannot reboot at this time as it is live and many people are using it. Maybe overnight, but likely not until the weekend.

 

[Chris Tobey]

Well, I made the mistake of trying to detach and auto-import the drive... It detached just fine, but the auto-import cannot find it.

 

[cyberjock]

Well, I made the mistake of trying to detach and auto-import the drive... It detached just fine, but the auto-import cannot find it.

That makes it sound like the replication not working out like expected is a symptom and not a cause of your problems. I would definitely do an upgrade to 9.2.1.9 though.

 

[Chris Tobey]

I will upgrade to 9.2.1.9 when I can get some downtime and then try this process again. Thanks cyberjock!

 

[Glorious1]

This could be way off base, but I notice your datasets are all at the root level of the pool. I was under the impression that user-created datasets should be under the automatically created root dataset of the pool?

Nevermind - I just read in another thread that the root dataset was hidden or mixed in the line with the pool in versions before 9.3.

 

[Chris Tobey]

Hey everyone, I was finally able to get this to work!

All I had to do was detach the pool and re-import it. I was on the right track, but when I tried this before it failed. I did not realize that I had enabled encryption on my SG2 pool until I went to remake it in 9.2.1.9 and there was a popup with some more information as I blindly clicked through. So my latest attempt at creating the RAIDZ-2 SG2 pool was made without using encryption, and everything worked as it should.

I do not know why the step of detaching and importing is necessary, but for now it seems it is and I can live with that.