Partially, but DB does not needed to pass audits. It is used when the piece requested for repair.
The only way to fail audit when your node is online and answered on audit request but does not able to provide a correct hash (three times each in a different time). Either disk was inaccessible or data is lost.
Wouldn’t this be a risk for repair if too many nodes had db issues but would still pass audits?
I believe it only tries three times in case of a time out. If the node can’t access the data or returns a bad hash, it’s an instant audit fail. But please correct me if I’m wrong.
yes, if it has a timeout or unknown error.
The audit considered as failed, if the returned hash doesn’t match.
Perhaps. I do not know, why the database is used for that specific request.