Checksum Inherit (off) (bug?)

[Darrell Patenaude]

Hello, recently installed Truenas Scale and I have noticed something strange.
Main Filesystem called Serverpool
I go into Edit Options, click advanced options, scroll down and select Checksum ON (which I assume this is for ZFS)
Then I create my dataset ie Programs
And I click edit options and advanced options and most of the time it says under checksum inherit (off)
I manually select ON, and save.

Then I'll come back later and it will either say ON still, or if I hover the mouse around there some times it changes to Inherit (off) in front of my eyes.

Can some one tell me what checksum means or does in the menu here, is it important.
Looking up checksum in google always just brings up many issues with people having checksum errors.

I also run snapshots and backup to a remote computer and also I backup to another set of drives on my main server.
And it almost always turns off checksum.

Any help or explaining if this is something I should be worrying about would be nice.

 

[Arwen]

From what I can tell, it's just an artifact of the menu. The option for Checksum has "Inherit (off)" as the first option. It's not that checksums are off. They should probably have the menu default to the existing configuration. You can submit a Jira bug on the problem.

The real test is to use the command line, and check the actual dataset configuration.


One of the key features of ZFS, is check summing of everything. Later, additional checksum algorithms were added to reduce potential issues for newer features like de-dup. So it is very unlikely that a NAS would actually disable ZFS checksums.

 

[zebekias]

I have to confess, that threw me off as well. All of my datasets have checksum as confirmed via zfs get checksum <dataset> but as Arwen mentioned, the GUI does not show the existing configuration, but it's just the first item on the list of possible values.

 

[neptunus]

I had the same today. Did someone submitted a Jira bug on the problem?

BTW. I had this issue on TrueNAS Core.

 

[winnielinnie]

A bug was already submitted over two months ago, and now it's been closed with the reason "Closed Without Changes. Cannot Reproduce."

This is strange, because I can still reproduce this on TrueNAS Core 13.0-U4.

This is 100% a bug with the TrueNAS GUI. Not a bug in ZFS.

So unless they re-open the bug, looks like it's going to stay.

[NAS-120203] - iXsystems TrueNAS Jira

ixsystems.atlassian.net

Here is the true information, using "zfs" commands in the terminal:

Code:

$ zfs get checksum mypool/playground
mypool/playground checksum on default

Here is the GUI giving incorrect information about the dataset's checksum property:

 

[neptunus]

A bug was already submitted over two months ago, and now it's been closed with the reason "Closed Without Changes. Cannot Reproduce."

This is strange, because I can still reproduce this on TrueNAS Core 13.0-U4.

This is 100% a bug with the TrueNAS GUI. Not a bug in ZFS.

So unless they re-open the bug, looks like it's going to stay.

[NAS-120203] - iXsystems TrueNAS Jira

ixsystems.atlassian.net

Here is the true information, using "zfs" commands in the terminal:

Code:

$ zfs get checksum mypool/playground
mypool/playground checksum on default

Here is the GUI giving incorrect information about the dataset's checksum property:
View attachment 66158

I see same like winnielinnie.

 

[santacruzskim]

Same result here in UI vs terminal on TrueNas Core. I caught the "Inherit (off)" listing one day and assumed I had done something stupid so I went back and manually flipped all 10 datasets to "ON," but when I went back to check they had flipped back to "Inherit (off)."

Luckily, between this thread and confirming in the terminal I'm able to breathe easy, but it's still frustrating and clearly a bug. This is literally a perfect example to use when explaining what a "bug" is to someone, and I'm reproducing it just fine :)