My knowledge is exhausted. Sorry.
I can try to start all over using backups to restore most data, but I almost lost my faith in TrueNas as a reliable platform.
I had a mirrorred setup and believed it to be stable and reliable using ZFS but that does not appear to be true. How can I make sure I never get in this situation again with TrueNas, without having to backup everything?
I used ‘zdb -ul’ to find out whether there was a transaction group (TXG) previous the corruption.
I found a good candidate and used ‘zpool import -f -F -T ’ to try to import this TXG.
It reported the pool was salvageable, so I continued with ‘zpool import’ to import the pool.
After some disk processing the system rebooted, but unfortunately it now hangs again after ‘loaded plugin middleware.plugins.disk’
The conclusion is that salvaging did not succeed, but I gained a lot of knowledge about ZFS. Will from scratch now using my backups.
Thanks for helping me on the way.
-
If you manage to succeed with an import, you mustn’t reboot until you have checked a few things.
-
You need to check that the pool and datasets have mounted in the correct location.
-
When you do stuff with the CLI, the UI can get out of step - so you need to export in the CLI and export in the UI and then import in the UI.
When it all looks OK, and you have exported and imported in the UI and it still looks OK, then you can reboot and check that it all loads correctly on boot.
You always need to do backups. No matter what redundancy you have.
Absolutely – redundancy is NOT a backup - and nor are snapshots.