diff --git a/client/network/src/protocol.rs b/client/network/src/protocol.rs
index bbb87b5255d10..1ae1d48cba3c5 100644
--- a/client/network/src/protocol.rs
+++ b/client/network/src/protocol.rs
@@ -526,7 +526,9 @@ impl<B: BlockT> Protocol<B> {
 		}
 
 		if let Some(_peer_data) = self.peers.remove(&peer) {
-			self.sync.peer_disconnected(&peer);
+			if let Some(sync::OnBlockData::Import(origin, blocks)) = self.sync.peer_disconnected(&peer) {
+				self.pending_messages.push_back(CustomMessageOutcome::BlockImport(origin, blocks));
+			}
 			Ok(())
 		} else {
 			Err(())
diff --git a/client/network/src/protocol/sync.rs b/client/network/src/protocol/sync.rs
index dd682bf348b03..6e07bd4c96971 100644
--- a/client/network/src/protocol/sync.rs
+++ b/client/network/src/protocol/sync.rs
@@ -738,10 +738,19 @@ impl<B: BlockT> ChainSync<B> {
 			// If our best queued is more than `MAX_BLOCKS_TO_LOOK_BACKWARDS` blocks away from the
 			// common number, the peer best number is higher than our best queued and the common
 			// number is smaller than the last finalized block number, we should do an ancestor
-			// search to find a better common block.
+			// search to find a better common block. If the queue is full we wait till all blocks are
+			// imported though.
 			if best_queued.saturating_sub(peer.common_number) > MAX_BLOCKS_TO_LOOK_BACKWARDS.into()
 				&& best_queued < peer.best_number && peer.common_number < last_finalized
+				&& queue.len() <= MAJOR_SYNC_BLOCKS.into()
 			{
+				trace!(
+					target: "sync",
+					"Peer {:?} common block {} too far behind of our best {}. Starting ancestry search.",
+					id,
+					peer.common_number,
+					best_queued,
+				);
 				let current = std::cmp::min(peer.best_number, best_queued);
 				peer.state = PeerSyncState::AncestorSearch {
 					current,
@@ -804,7 +813,7 @@ impl<B: BlockT> ChainSync<B> {
 		response: BlockResponse<B>
 	) -> Result<OnBlockData<B>, BadPeer> {
 		self.downloaded_blocks += response.blocks.len();
-		let mut new_blocks: Vec<IncomingBlock<B>> =
+		let new_blocks: Vec<IncomingBlock<B>> =
 			if let Some(peer) = self.peers.get_mut(who) {
 				let mut blocks = response.blocks;
 				if request.as_ref().map_or(false, |r| r.direction == message::Direction::Descending) {
@@ -970,6 +979,13 @@ impl<B: BlockT> ChainSync<B> {
 				return Err(BadPeer(who.clone(), rep::NOT_REQUESTED));
 			};
 
+		Ok(self.validate_and_queue_blocks(new_blocks))
+	}
+
+	fn validate_and_queue_blocks(
+		&mut self,
+		mut new_blocks: Vec<IncomingBlock<B>>,
+	) -> OnBlockData<B> {
 		let orig_len = new_blocks.len();
 		new_blocks.retain(|b| !self.queue_blocks.contains(&b.hash));
 		if new_blocks.len() != orig_len {
@@ -992,10 +1008,8 @@ impl<B: BlockT> ChainSync<B> {
 			);
 			self.on_block_queued(h, n)
 		}
-
 		self.queue_blocks.extend(new_blocks.iter().map(|b| b.hash));
-
-		Ok(OnBlockData::Import(origin, new_blocks))
+		OnBlockData::Import(origin, new_blocks)
 	}
 
 	/// Handle a response from the remote to a justification request that we made.
@@ -1353,7 +1367,7 @@ impl<B: BlockT> ChainSync<B> {
 					PreValidateBlockAnnounce::Failure { who, disconnect }
 				}
 				Err(e) => {
-					error!(
+					debug!(
 						target: "sync",
 						"💔 Block announcement validation of block {:?} errored: {}",
 						hash,
@@ -1542,11 +1556,34 @@ impl<B: BlockT> ChainSync<B> {
 	}
 
 	/// Call when a peer has disconnected.
-	pub fn peer_disconnected(&mut self, who: &PeerId) {
+	/// Canceled obsolete block request may result in some blocks being ready for
+	/// import, so this functions checks for such blocks and returns them.
+	pub fn peer_disconnected(&mut self, who: &PeerId) -> Option<OnBlockData<B>> {
 		self.blocks.clear_peer_download(who);
 		self.peers.remove(who);
 		self.extra_justifications.peer_disconnected(who);
 		self.pending_requests.set_all();
+		let blocks: Vec<_> = self.blocks
+			.drain(self.best_queued_number + One::one())
+			.into_iter()
+			.map(|block_data| {
+				let justifications =
+					legacy_justification_mapping(block_data.block.justification);
+				IncomingBlock {
+					hash: block_data.block.hash,
+					header: block_data.block.header,
+					body: block_data.block.body,
+					justifications,
+					origin: block_data.origin,
+					allow_missing_state: true,
+					import_existing: false,
+				}
+			}).collect();
+		if !blocks.is_empty() {
+			Some(self.validate_and_queue_blocks(blocks))
+		} else {
+			None
+		}
 	}
 
 	/// Restart the sync process. This will reset all pending block requests and return an iterator
@@ -2349,6 +2386,9 @@ mod test {
 					.for_each(|b| block_on(client.import_as_final(BlockOrigin::Own, b)).unwrap());
 		}
 
+		// "Wait" for the queue to clear
+		sync.queue_blocks.clear();
+
 		// Let peer2 announce that it finished syncing
 		send_block_announce(best_block.header().clone(), &peer_id2, &mut sync);