It is the only thing It would be…. Scrubs cant repair It. Zpool status show permanente errors, but not list of affected files. I cant remove some corrupted directories due to corruption…
Any ideas to try to recover pool??
AFAIK, no app or read or write operation should be able to cause checksum errors. Or kill metadata.
I think my English is not good enough…. The reason of checksum error increase rate was the app trying to read conrrupted data app! Not the reason of corruption obviously!
Once the app stop, the chechsum errors stop, beacuse nothing is trying to read the data corrupted.
So, my checksum errors only appear when someone or apps trying to read corrupted files…
But you suspect the errors are caused by the app trying to read the corrupted subfolder. So can’t you sync only that from your main NAS and test if the scrub succeeds then? You should be able to delete the corrupted folder with rm -rf and you would also need to delete all the snapshots of that dataset so that the scrub doesn’t detect it again
Carefull now…
You dont want him to execute that in the wrong folder.
no way! rm -rf dont work too. Output for that comand says:
rm: cannot remove ‘/mnt/Poll 1 TB/FotoVideo_Hogar/Originales_Hogar/Immich’: Input/output error
This one is another affected directory…!!! 
Could it be related to the space character in the dataset/folder name?
So, my checksum errors only appear when someone or apps trying to read corrupted files…
Sure, or when you run a scrub.
But that is not the App causing the issue, that is the app forcing the fs to find the issue. Just like a scrub would.
To find out what is causing the issue, you have to swap stuff and see if the issues persist.
I lost track of what you swapped and what not.
HDD
SATA data cables
SATA power cables
PSU
CPU
RAM
Mobo
LSI or onboard SATA