Fast sync download fix on restart #175
Conversation
| requestedMptNodes = requestedMptNodes - ref | ||
| requestedNonMptNodes = requestedNonMptNodes - ref | ||
| requestedBlockBodies = requestedBlockBodies - ref | ||
| requestedReceipts = requestedReceipts - ref |
There was a problem hiding this comment.
I'm wondering, should we add any request being handled by the terminated handler to our queues of pending requests? For example if a FastSyncNodesRequestHandler terminated we may have to add the nodes it was in charge of to mptNodesQueue and nonMptNodesQueue lists
There was a problem hiding this comment.
it's done in the FastSyncNodesRequestHandler itself:
override def handleTerminated(): Unit = {
syncController ! FastSync.EnqueueNodes(requestedHashes)
cleanupAndStop()
}
There was a problem hiding this comment.
hmm actually that's what happens when underlying peer actor is terminated, not the handler
| requestedMptNodes = requestedMptNodes - ref | ||
| requestedNonMptNodes = requestedNonMptNodes - ref | ||
| requestedBlockBodies = requestedBlockBodies - ref | ||
| requestedReceipts = requestedReceipts - ref |
There was a problem hiding this comment.
Terminated message will happen after requested entities will be added back to queue?
There was a problem hiding this comment.
let me double check
There was a problem hiding this comment.
seems you're right @adamsmo we should enqueue these items here
This fixes the issue when, after restarting during fast sync, some mpt nodes were "lost". Solution: we need to also save requested items in sync state (the ones that are not yet in db, but not in queues anymore).