TrueNAS SCALE 24.04-BETA.1 (Dragonfish) has been released!

[Davvo]

Thank you for your work. I was wondering about the following:

• An unsupported development mode is added to the base system. Enabling this puts the system in an unsupported state and allows customization of the operating system.

Iirc I read sometime ago that doing so would enable a flag in the big report to ease your work. Is this confirmed or does this unsupported state mean more?

 

[joeschmuck]

Thank you for your work. I was wondering about the following:

Iirc I read sometime ago that doing so would enable a flag in the big report to ease your work. Is this confirmed or does this unsupported state mean more?

I'm curious to what you are asking. I have used the developer tools, it basically unlocks TrueNAS to work like the Debian OS. It can be very handy if you are trying to do some development. I'm glad they included this feature to TrueNAS SCALE.

 

[morganL]

Thank you for your work. I was wondering about the following:

Iirc I read sometime ago that doing so would enable a flag in the big report to ease your work. Is this confirmed or does this unsupported state mean more?

It means we've moved the restrictions on apt-get etc, but also can track this in debug. It's unsupported and intended for developers and not "production".

If a user finds a bug in the system... we'd prefer you turn off this mode and replicate the bug.

Software Updates will not keep the "developer changes". So, best to script and re-install changes.

 

[Jorsher]

Have you tried making bug reports from the UI in Dragonfish?

I have not. I didn't mean for that comment to be taken seriously. :)

Being able to provide feedback directly from the UI is definitely welcome, just haven't had a reason to use it yet.

 

[CJRoss]

That is very true for older NVMe drives. I've found some that provide very little information. The new NVMe drives coming out "should" support he current NVMe specifications and the data will be there, hopefully.

What do you consider newer? I was just testing smart tools on the 980 Pro in my desktop and while it provides a lot of information, I wasn't able to get it to run a test nor report back on any tests. I think it was released in 2020 so I'm curious what you think the cutoff is.

 

[joeschmuck]

What do you consider newer? I was just testing smart tools on the 980 Pro in my desktop and while it provides a lot of information, I wasn't able to get it to run a test nor report back on any tests. I think it was released in 2020 so I'm curious what you think the cutoff is.

NVMe Specification 1.3 which came out in May 2017 includes SMART Self-Test (Optional), Spec 1.4 from June 2019 is also Optional, however Spec 2.0 from Dec 2023 does not list it as optional anymore.

In SCALE (Debian) run these CLI commands:
nvme id-ctrl /dev/nvme0 -H | grep -i "self-test" and it should tell if SMART is or is Not supported. But I know the 980 PRO does support SMART already. I do not know yet if it supports self-test, maybe you can tell me.

If you want to run a self-test, if 'smartctl' is not doing it (needs to be at least version 7.4 but NVMe improvements are coming in v7.5) then you can do the following:

Making an assumption that nvme0 is the drive you desire to use.

nvme device-self-test /dev/nvme0 -s 1 for a SHORT test. Use nvme device-self-test /dev/nvme0 -s 2 for a LONG test.
nvme self-test-log --output=json /dev/nvme0 will dump the test results. There is a little bit of decoding needed.

Decoding:
Self test result: 0=Pass, 1=Fail
Self test code: 1 = Short, 2 = Long

 

[CJRoss]

NVMe Specification 1.3 which came out in May 2017 includes SMART Self-Test (Optional), Spec 1.4 from June 2019 is also Optional, however Spec 2.0 from Dec 2023 does not list it as optional anymore.

In SCALE (Debian) run these CLI commands:
nvme id-ctrl /dev/nvme0 -H | grep -i "self-test" and it should tell if SMART is or is Not supported. But I know the 980 PRO does support SMART already. I do not know yet if it supports self-test, maybe you can tell me.

If you want to run a self-test, if 'smartctl' is not doing it (needs to be at least version 7.4 but NVMe improvements are coming in v7.5) then you can do the following:

Making an assumption that nvme0 is the drive you desire to use.

nvme device-self-test /dev/nvme0 -s 1 for a SHORT test. Use nvme device-self-test /dev/nvme0 -s 2 for a LONG test.
nvme self-test-log --output=json /dev/nvme0 will dump the test results. There is a little bit of decoding needed.

Decoding:
Self test result: 0=Pass, 1=Fail
Self test code: 1 = Short, 2 = Long

I currently have smartctl 7.3 so that appears to be why I don't have any luck with it. Also, I think the last time I played with NVMe SMART testing was with my Gigabyte drive, although that one came out in 2019.

I did try the self-test previously when I was testing something with nvme-cli but it didn't actually run a test. Looking at my history it looks like I had the arguments wrong. I was just able to complete a short self test and it passed. Also, the json argument isn't really needed unless you care about the power on hours.

And Core only has smartctl 7.2 with Scale at 7.3(unless Dragonfish includes a newer version), so it's all academic ATM.

 

[joeschmuck]

Also, the json argument isn't really needed unless you care about the power on hours.

This makes it more human readable and of course I always reference the power_on_hours, that is the only way you really know the last test was not done 2 months ago.

 

[morganL]

FYI - there are over 1,000 Dragonfish.BETA deployments now. It's ahead of normal pace.

There have been about 20 minor issues... to be fixed in RC.1 (mid-March).

So, progress is good and we're at a better state than normal. Appreciate any additional feedback and bug reports.