What
Because a node is created in the DB, then added to a shard in a callback, if something goes wrong during adding of the node, it can end up in an orphaned state -- with no node_in_shards record and nothing left to notice that.
Adding nodes to shards has become a more complicated process that is potentially error prone (such as having to catch up a pieceCID calculation cause the shard is in a bad state), so we need to account for this
What
Because a node is created in the DB, then added to a shard in a callback, if something goes wrong during adding of the node, it can end up in an orphaned state -- with no node_in_shards record and nothing left to notice that.
Adding nodes to shards has become a more complicated process that is potentially error prone (such as having to catch up a pieceCID calculation cause the shard is in a bad state), so we need to account for this