Datasets missing in webUI

[o1982]

Hi All,

Yet another problem with SCALE - my Datasets are missing in the web UI. I can see the disks under Storage>"pool name">Manage Devices and all of the disks are there with "No errors". There are green ticks across the board for my two pools. Using "mc" in the shell I can see the datasets (and child datasets) under /mnt/. Is there any way to get them back? The SMB share is also a bit temperamental - it takes some time to show files and folders in a share.

Any help appreciated.

FML.

 

[sretalla]

Sounds like you may have manually imported your pool at the CLI?

In that case, export at the CLI and import again in the GUI.

 

[o1982]

Thanks for the tip.

No, I did not do anything with the pool through the cli.

 

[sretalla]

OK, so I'm not sure I get what you're seeing then.

Can you share a screenshot or two to illustrate the problem?

 

[Arwen]

I would also suggest the output of zfs list -r in addition to the GUI snapshot. That will allow us to compare what ZFS datasets exist and should be in the GUI.

 

[o1982]

Thank you both, I'll take another look at it this morning...

 

[o1982]

Ok, so currently in the webui I see no Pools or Datasets. SMB shares are still listed, but disabled.

Using the webui shell and "zfs list -r" I see all of the parent and child datasets that I would expect to see listed in the webui...but they are not. I tried a simple restart to really refresh the webui in case it was simply out of sync somehow, but there were no changes.

I can export and import using the cli but the webui doesn't list anything to import on the storage tab.

In that case, export at the CLI and import again in the GUI.

@sretalla - do you mean the normal linux shell or the TrueNAS CLI (through the web? or is local console better?)

 

[sretalla]

normal linux shell

That should do it... the TrueNAS CLI is actually calling the same APIs as the GUI, so potentially won't do it.

zpool export poolname at the shell, then use the GUI to import it again.

 

[o1982]

zpool export poolname at the shell, then use the GUI to import it again.

It worked for my "Data" Pool, datasets were shown in the webui for it, but when I tried it for the "Services" Pool it did not. I repeated "zpool export Services (which completed without error) and imported it again in the webui, but the webui still didn't show the Datasets for the Services pool. I restarted and now ALL the datasets are missing again in the webui.

 

[sretalla]

That all sounds very concerning... do you see any errors in the output of dmesg or in /var/log/middlewared.log ?

 

[o1982]

Hi @sretalla , thank you for your help.

dmesg was ok, but middlewared.log has lots of errors with regard to dataset/pool operations. I see alot of:

RuntimeError: (1, 'Corrupt database')

Is this easy to fix? If not, I have downloaded the config file - would it be easy to restore my config on a new install and import the pools again? Or would I be copying a corrupt database or something that is causing the DB to become corrupt on a new install?

At this point (considering the other problems I have had with SCALE (Network stupidity ("two NICs on the same subnet? Are you mad?! And you want a msg in the webui for that? Ha!"), App routes/VLANs, PCI Passthrough problems, SMB & App Permissions (we're supposed to setup an NFS Share for the app? That's totally riduculous) and other GUI weirdness...I think if I can't get it working reliably today I will switch to Ubuntu server or Proxmox. I might come back when the beta testing is really complete.

 

[sretalla]

RuntimeError: (1, 'Corrupt database')

Is this easy to fix?

OK, so that's a tricky one as your config is effectively that database and hence restoring the problem will just bring the problem back (maybe... depending on what exactly is corrupt).

There's probably not a lot to be lost by trying that though, so worst case, it doesn't work, but you're just in the same place.

If that's how it works out, a "fresh install" is the likely only road forward... re-do your config after re-import of the pools.

At this point (considering the other problems I have had with SCALE (Network stupidity ("two NICs on the same subnet? Are you mad?! And you want a msg in the webui for that? Ha!"), App routes/VLANs, PCI Passthrough problems, SMB & App Permissions (we're supposed to setup an NFS Share for the app? That's totally riduculous) and other GUI weirdness...I think if I can't get it working reliably today I will switch to Ubuntu server or Proxmox. I might come back when the beta testing is really complete.

Although a lot of what I see there is somewhat common, most of the problem is just a learning curve and the database corruption issue is very rare (not even sure I've seen it twice in 6 years).

It is probably worth the trouble of the learning curve, but the decision is yours to make.

 

[pwapwa]

It appears that I have the same problem. Running TrueNAS Scale 22.12.2
The system has been stable for quite a while. About a week ago I noticed that when trying to display Datasets on the Web GUI I get

"No Datasets​

It seems you haven't configured pools yet. Please click the button below to create a pool."

If I list Storage -> Storage Dashboard shows all pools as expected.
Other than that the system appears to work well.

Results of cat middlewared.log | grep Error | grep dataset yields
middlewared.service_exception.CallError: [EZFS_NOENT] Failed to import 'Temp-Mirrored' pool: cannot import 'Temp-Mirrored' as 'Temp-Mirrored': no such pool or dataset
middlewared.service_exception.CallError: [EZFS_NOENT] Failed to import 'Temp-Mirrored' pool: cannot import 'Temp-Mirrored' as 'Temp-Mirrored': no such pool or dataset
middlewared.service_exception.CallError: [EZFS_NOENT] Failed to import 'RaidMaster-RaidZ1' pool: cannot import 'RaidMaster-RaidZ1' as 'RaidMaster-RaidZ1': no such pool or dataset
middlewared.service_exception.CallError: [EZFS_NOENT] Failed to import 'RaidMaster-RaidZ1' pool: cannot import 'RaidMaster-RaidZ1' as 'RaidMaster-RaidZ1': no such pool or dataset
middlewared.service_exception.ValidationErrors: [EINVAL] chart_release_create.volumes.0: Invalid dataset name /ps/library. "test1, ix-test, ix_test" are valid examples.

The pools / datasets show up in storage and the storage->disks displays as expected.
cat middlewared.log | grep corrupt and cat middlewared.log | grep Corrupt
yield nothing in log, log.1, log.2 and log.3

dmesg has no obvious errors concerning pools and datasets. It does complain about Duplicate Cookie Detected. But that
Apart from no datasets appearing, the system appears to be normal and operating as expected.

I can go to Temp-Mirrored in the shell and it seems to be fine.

What should / can I do to get datasets to show and resolve whatever is causing a problem?