ZFS file system again, but this time with usage experience.

[jabbr]

17 hours ago, AudioDoctor said:

 

yes I am aware of that, nor did I say that. I said that the system uses them to compare as well as the mirror or parity drives to decide which is correct.

the system doesn't use snapshots to compare. it doesn't maintain full copies of old files, rather just the changed blocks.

[AudioDoctor]

2 hours ago, jabbr said:

the system doesn't use snapshots to compare. it doesn't maintain full copies of old files, rather just the changed blocks.

 

In my reading, that was what was said.

[AudioDoctor]

SO answer the question, In a two disk mirror, how does the system know that the mirror is correct rather than the main drive? What if the parity drive is the drive with the error? How does it know which of the two versions is correct when it finds an error?

[jabbr]

10 minutes ago, AudioDoctor said:

SO answer the question, In a two disk mirror, how does the system know that the mirror is correct rather than the main drive? What if the parity drive is the drive with the error? How does it know which of the two versions is correct when it finds an error?

From the source: https://blogs.oracle.com/bonwick/zfs-end-to-end-data-integrity

There's a ton more that's been written

[AudioDoctor]

11 minutes ago, jabbr said:

From the source: https://blogs.oracle.com/bonwick/zfs-end-to-end-data-integrity

There's a ton more that's been written

 

ok, well, I stand corrected and I will eliminate that from my pool of ZFS knowledge.

[Jud]

I see that to really use ZFS as intended, it ought to be installed on all system storage, rather than our old notion of installing on a partition.  So I'd best save experimentation with it for building NAS, whenever I get around to that.

[jabbr]

9 minutes ago, Jud said:

I see that to really use ZFS as intended, it ought to be installed on all system storage, rather than our old notion of installing on a partition.  So I'd best save experimentation with it for building NAS, whenever I get around to that.

Yes absolutely and it likes ECC memory. My "NAS" is a low powered Linux server which has 32Gb ECC RAM and which runs Ubuntu linux. Mine is connected to a 15bay SAS3 enclosure but that doesn't matter.

[jabbr]

43 minutes ago, AudioDoctor said:

 

ok, well, I stand corrected and I will eliminate that from my pool of ZFS knowledge.

The short answer is that the chance that the stored hash for that block would be corrupted to match the corrupted block is extremely small ... in that case the system would still know that the blocks don't match but would be unable to repair ... and the point is that ZFS isn't a backup either ... so store important data offsight as well ... actually you can automatically store multiple copies of files but again if the array burns down you are out of luck unless there is an external backup.

 

Most people can simply mirror drives and ZFS really just works...

[Jud]

11 minutes ago, jabbr said:

The short answer is that the chance that the stored hash for that block would be corrupted to match the corrupted block is extremely small ... in that case the system would still know that the blocks don't match but would be unable to repair ... and the point is that ZFS isn't a backup either ... so store important data offsight as well ... actually you can automatically store multiple copies of files but again if the array burns down you are out of luck unless there is an external backup.

 

Most people can simply mirror drives and ZFS really just works...

 

@AudioDoctor's brother has a similar setup, I believe, so his chance of losing data to corruption is astronomically smaller, excepting some sort of natural or manmade disaster widescale enough to take out both installations.

[AudioDoctor]

I had my first, operator induced, ZFS breakdown today and with a few quick commands was back up and running from a previous snapshot of the boot drive. A quick sudo apt install -f to get some dependencies and make sure everything was ship shape and we're off to the races.

 

 

On 5/16/2021 at 4:54 PM, Jud said:

 

@AudioDoctor's brother has a similar setup, I believe, so his chance of losing data to corruption is astronomically smaller, excepting some sort of natural or manmade disaster widescale enough to take out both installations.

 

Yeah, we're actually transitioning to a ZFS based backup system and trying to figure out how to keep the fire and water proof systems in the mix somehow.

[AudioDoctor]

And after the repair, are all my files okay?

 

[jabbr]

On 5/29/2021 at 7:20 PM, AudioDoctor said:

I had my first, operator induced, ZFS breakdown today and with a few quick commands was back up and running from a previous snapshot of the boot drive. A quick sudo apt install -f to get some dependencies and make sure everything was ship shape and we're off to the races.

 

 

 

Yeah, we're actually transitioning to a ZFS based backup system and trying to figure out how to keep the fire and water proof systems in the mix somehow.

I don't have enough info to answer.

 

Do a 

$ zpool status

and if no errors, then your files should be a-ok

 

if you are missing a pool then $ zpool import -a 

 

[ note that if you install Ubuntu with the Zsys/ZFS root feature, then the system does a snapshot before each "apt install" so that you can rollback ]

[AudioDoctor]

1 hour ago, jabbr said:

I don't have enough info to answer.

 

Do a 

$ zpool status

and if no errors, then your files should be a-ok

 

if you are missing a pool then $ zpool import -a 

 

[ note that if you install Ubuntu with the Zsys/ZFS root feature, then the system does a snapshot before each "apt install" so that you can rollback ]

thanks. Everything is running fine after my mistake.