Volume degraded ? until serverrestart then online again for some time ...

[Sir.Robin]

Errh... oh man... :(

Why did you rollback? What were you trying to accomplish with that?

This is a snapshot taken on the volume level? Not a Dataset?

Did you send the data from volume01 to volume02 or volume02/somedataset ?

No backup?

 

[arameen]

Well what I tried to do was to copy the contents of pool 1 to pool 2, so i can recreate pool 1 as a raidz2 instead of raidz3. Before i can destroy pool 1 I needed to back it up and did choose to do that on pool 2 where i had lots of storage. I didnt know that this process will substitute pool 2 with pool 1 contents, thought I will just add pool 1 content to pool 2 existing data.
No, I have no complete backup of pool 2 contents, only of some folders there.
If I at least knew what I lost then it will be easier, already thinking that those folders are gone.
A list of the contents in freenas?
Or if the CICS share somehow stored the tree structure of the files so I could read what was in there.

 

[Darren Myers]

I JUST posted about copying data from 1 dataset on a pool to another :( sorry man wish i could help.

 

[Sir.Robin]

Well what I tried to do was to copy the contents of pool 1 to pool 2, so i can recreate pool 1 as a raidz2 instead of raidz3. Before i can destroy pool 1 I needed to back it up and did choose to do that on pool 2 where i had lots of storage. I didnt know that this process will substitute pool 2 with pool 1 contents, thought I will just add pool 1 content to pool 2 existing data.
No, I have no complete backup of pool 2 contents, only of some folders there.
If I at least knew what I lost then it will be easier, already thinking that those folders are gone.
A list of the contents in freenas?
Or if the CICS share somehow stored the tree structure of the files so I could read what was in there.

Yes i got that... but after you pick the most recent snap and send it to the second pool... why did you do a rollback? There would be no need for it as far as i can tell.

You could and should send everything to it's own dataset on the second pool.

 

[arameen]

just wanted to make sure i had a copy of pool 1, as new to freeNAS I could only veryfiy this by seeing the files, that needed a rollback.
Done is done
Question is if there is any way to revert this operation, undo it, restore the files or at least get a treescruture of what is gone. As mentioned, i dont have any snapshot of the second pool.

 

[Sir.Robin]

You sure it is actually ghone? Have you done a "zfs list" on the cli?

 

[arameen]

well not 100%, I am a newbie after all :p

Main_Raidz3 is pool 1 while Secondary_Raidz2 is pool 2 (the one that lost its data and now contains the other pools data)

But it doesn't look like its there anymore!
Unless some log or something else makes it possible to restore, revert or undo this rollback of snapshot so the second pool looks like it did before

 

[Sir.Robin]

how about a "zfs list -t snapshot"?

No datasets on either one except for the rwar? Did you create a CIFS share directly on the volume?

 

[arameen]

yes i have 2 CIFS shares, one on each pools dataset.
And no snapshots except the one i created for pool 1 and moved to pool 2 for rollback.

 

[Sir.Robin]

Right... so you used zfs send/receive to "copy" your data from one pool to the other.

This is not a "copy data job" like "copy/paste" in windows explorer. You are replicating a filesystem from one pool/volume to another one.

If i'm not entirely wrong, this means that you overwrote your Secondary_Raidz2 already when sending your snapshot.

Either way, i don't think you will be able to get your data back.

 

[arameen]

So far, i figured out that, once i looked into the second pool :(

Question is, if there is anything to do?
Undo, revert, back this rollback of snapshot process ?

 

[Sir.Robin]

No, sorry. Even if the Secondary_Raidz2 pool had snapshots... neither would they exsist anymore.

Gonna do a test i a VM... to be sure.

 

[Sir.Robin]

yes i have 2 CIFS shares, one on each pools dataset.
And no snapshots except the one i created for pool 1 and moved to pool 2 for rollback.

So if i understand you correctly. You are misunderstanding some things.
You created a snap for "rollback". That does not mean you can undo your changes. It means you roll back your filesystem to that snapshot, that point in time. And the snapshot is on your pool1. No good for your pool2.

Test:

I now have two pools. vol01 and vol02. Both has their files and folders.

In addition:
Vol01 has the .system dataset
Vol02 has a empty dataset

I create a snap01 on vol01. Non recursive (not doing descendent file systems ie. Datasets).

This snapshot i want to send like you did to my vol02 for backup.

Code:

zfs send vol01@snap01 | vol02

And a error arrives: cannot receive new filesystem stream: destination 'vol02' exists must specify -F to overwrite it

Did you use -F ? I bet u did... and it clearly says what it does: -F to overwrite it.

So i do this to test.

Code:

zfs send vol01@snap01 | zfs receive -F vol02

And voila. All files/folders on the vol02 are replaced by the contents of vol01. The filesystem is overwritten.
The only thing left is my dataset on vol02 because it is a different filesystem and the .System is not replicated because i did not do a recursive snapshot.

So... not much help... but at least an explanation to what you did wrong...

You would be home free if you sent your snap to its own dataset on pool2.

Use datasets for everything.

 

[arameen]

Yes you understood.
the snapshot of pool 1 in pool 2 was intended to give me a backup of pool1 before destroying it.
Yes i used the F flag, actually i just used a command that someone suggested for me on another thread. Now i know what that f stands for
Only reason i choosed to use the command was something that didnt work in the GUI in that thread.

Anyway, thanks for the help and experiment. I should set up a freenas virtual machine to now on experiment on before doing some stuff. Even if that wouldnt have helped me to avoid overwriting because i didnt konw the F flag. In the end it is about not finding enough of information on how to do things in freenas.

 

[arameen]

Can you get smart data for your drives? smartctl -a /dev/adaX

I also suspect your raid card is what is causing the problems.

Well it is not, any suggestions who to move on :) ?

 

[arameen]

The ultra-crummy SATA card. Get an SAS expander or an additional LSI SAS 2008/2308 controller.

the cables are ok, the SATA card is not the problem either.

 

[SweetAndLow]

the cables are ok, the SATA card is not the problem either.

How do you know it's not the problem?

 

[arameen]

What happend now was that I got the critical message again saying:
CRITICAL: The volume Secondary_Raidz2 (ZFS) state is ONLINE: One or more devices has experienced an unrecoverable error. An attempt was made to correct the error. Applications are unaffected.

What we know for sure now is.
- The SATA card is not the problem (I removed it from my case totally)
- The cable is not the problem (replaced the cable)
- One harddrive is not showing its serial number under view disks, unlike the rest of the disks.

 

[rogerh]

The serial number issue sounds a bit like https://bugs.freenas.org/issues/5418

Don't know if it is a separate issue from the pool being degraded, though.

Edit: am I right to assume that the pool now complaining of being degraded is not the same pool as the one you have subtracted a disk from in order to remove the SATA card?

 

[Bidule0hm]

Same for me for the S/N not showing up, it's a well-known bug and it doesn't harm the pool so your problem is elsewhere.