Consensus system downtime recovery

1. A computer-implemented consensus method to be implemented on a blockchain maintained by a number (N) of nodes, wherein one of the nodes acts as a primary node and the other (N−1) nodes act as backup nodes, and the method is performed by one of the backup nodes, the method comprising:

obtaining a pre-prepare message from the primary node;

multicasting a prepare message to at least some of the primary node and the other (N−2) backup nodes, the prepare message indicating an acceptance of the pre-prepare message;

obtaining (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes, wherein Q (quorum) is (N+F+1)/2 rounded up to the nearest integer, and F is (N−1)/3 rounded down to the nearest integer;

storing at least a minimal amount of consensus messages for recovery after one or more of the N nodes crash, wherein the minimal amount of consensus messages comprise the pre-prepare message and at least (Q−1) of the (Q−1) or more prepare messages;

multicasting a commit message to at least some of the primary node and the other backup nodes, the commit message indicating that the one backup node agrees to the (Q−1) or more prepare messages; and

obtaining, respectively from Q or more nodes among the primary node and the backup nodes, Q or more commit messages each indicating that the corresponding node a corresponding node of the Q or more nodes agrees to (Q−1) or more prepare messages received by the corresponding node.

2. The method of claim 1 , wherein:

before obtaining the pre-prepare message from the primary node, the method further comprises obtaining one or more transaction requests from at least one of: a client, the primary node, or one or more of the other backup nodes;

the pre-prepare message comprises an order of one or more transactions corresponding to the one or more transaction requests;

the multicast commit message indicates that the one backup node agrees to the order; and

the method further comprises packing the one or more transactions into a local copy of the blockchain maintained by the one backup node according to the order.

3. The method of claim 1 , wherein the minimal amount of consensus messages comprise:

only the pre-prepare message and the (Q−1) prepare messages.

4. The method of claim 1 , wherein:

the (Q−1) or more prepare messages include the multicast prepare message; and

the Q or more commit messages include the multicast commit message.

5. The method of claim 1 , after multicasting the commit message, further comprising:

performing a system restart; and

loading at least the stored minimal amount of consensus messages.

6. The method of claim 1 , after storing at least the minimal amount of consensus messages and before multicasting the commit message, further comprising:

performing a system restart; and

loading at least the stored minimal amount of consensus messages, wherein the loading comprises loading the pre-prepare message and the at least (Q−1) of the (Q−1) or more prepare messages.

7. The method of claim 6 , after storing at least the minimal amount of consensus messages and before multicasting the commit message, further comprising:

multicasting a view change message comprising the loaded pre-prepare message and the loaded at least (Q−1) of the (Q−1) or more prepare messages.

8. The method of claim 7 , after storing at least the minimal amount of consensus messages and before multicasting the commit message, further comprising:

obtaining from a new primary node a new view message indicating that the new primary node has received Q or more view change messages each indicating that a node that sent one of the Q or more view change messages agrees to the view change message;

multicasting another prepare message to at least some of the backup nodes including the new primary node, the another prepare message indicating an acceptance of the new view message; and

obtaining another (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes.

9. The method of claim 7 , after storing at least the minimal amount of consensus messages and before multicasting the commit message, further comprising:

obtaining, respectively each node of Q or more of the backup nodes, a view change message indicating that the each node of Q or more of the backup nodes agrees to the view change message;

multicasting to at least some of the backup nodes a new view message indicating that the one backup node acting as a new primary node has received the Q or more view change messages; and

obtaining another (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes.

10. The method of claim 6 , wherein performing the system restart comprises:

performing the system restart without triggering a view change.

11. A consensus system for maintaining a blockchain, wherein a number of N nodes maintain the blockchain with one of the N nodes acting as a primary node and the other (N−1) nodes acting as backup nodes, the consensus system acting as one of the (N−1) backup nodes and comprising one or more processors and one or more non-transitory computer-readable memories coupled to the one or more processors and configured with instructions executable by the one or more processors to cause the system to perform operations comprising:

obtaining a pre-prepare message from the primary node;

multicasting a prepare message to at least some of the primary node and the other (N−2) backup nodes, the prepare message indicating an acceptance of the pre-prepare message;

obtaining (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes, wherein Q (quorum) is (N+F+1)/2 rounded up to the nearest integer, and F is (N−1)/3 rounded down to the nearest integer;

storing at least a minimal amount of consensus messages for recovery after one or more of the N nodes crash, wherein the minimal amount of consensus messages comprise the pre-prepare message and at least (Q−1) of the (Q−1) or more prepare messages;

multicasting a commit message to at least some of the primary node and the other backup nodes, the commit message indicating that the one backup node agrees to the (Q−1) or more prepare messages; and

obtaining, respectively from Q or more nodes among the primary node and the backup nodes, Q or more commit messages each indicating that a corresponding node of the Q or more nodes agrees to (Q−1) or more prepare messages received by the node.

12. The system of claim 11 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

performing a system restart; and

loading at least the stored minimal amount of consensus messages, wherein the loading comprises loading the pre-prepare message and the at least (Q−1) of the (Q−1) or more prepare messages.

13. The system of claim 12 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

multicasting a view change message comprising the loaded pre-prepare message and the loaded at least (Q−1) of the (Q−1) or more prepare messages.

14. The system of claim 13 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

obtaining from a new primary node a new view message indicating that the new primary node has received Q or more view change messages each indicating that a node that sent one of the Q or more view change messages agrees to the view change message;

multicasting another prepare message indicating an acceptance of the new view message to at least some of the backup nodes including the new primary node; and

obtaining another (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes.

15. The system of claim 13 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

obtaining, from each node of Q or more of the backup nodes, a view change message indicating that the each node of Q or more of the backup nodes agrees to the view change message;

multicasting to at least some of the backup nodes a new view message indicating that the one backup node acting as a new primary node has received the Q or more view change messages; and

obtaining another (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes.

16. A non-transitory computer-readable storage medium for maintaining a blockchain, wherein a number of N nodes maintain the blockchain with one of the N nodes acting as a primary node and the other (N−1) nodes acting as backup nodes, the storage medium being associated with one of the (N−1) backup nodes and configured with instructions executable by one or more processors to cause the one or more processors to perform operations comprising:

obtaining a pre-prepare message from the primary node;

multicasting a prepare message to at least some of the primary node and the other (N−2) backup nodes, the prepare message indicating an acceptance of the pre-prepare message;

obtaining (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes, wherein Q (quorum) is (N+F+1)/2 rounded up to the nearest integer, and F is (N−1)/3 rounded down to the nearest integer;

storing at least a minimal amount of consensus messages for recovery after one or more of the N nodes crash, wherein the minimal amount of consensus messages comprise the pre-prepare message and at least (Q−1) of the (Q−1) or more prepare messages;

multicasting a commit message to at least some of the primary node and the other backup nodes, the commit message indicating that the one backup node agrees to the (Q−1) or more prepare messages; and

obtaining, respectively from Q or more nodes among the primary node and the backup nodes, Q or more commit messages each indicating that a corresponding node of the Q or more nodes agrees to (Q−1) or more prepare messages received by the corresponding node.

17. The storage medium of claim 16 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

performing a system restart; and

loading at least the stored minimal amount of consensus messages, wherein the loading comprises loading the pre-prepare message and the at least (Q−1) of the (Q−1) or more prepare messages.

18. The storage medium of claim 17 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

multicasting a view change message comprising the loaded pre-prepare message and the loaded at least (Q−1) of the (Q−1) or more prepare messages.

19. The storage medium of claim 18 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

obtaining from a new primary node a new view message indicating that the new primary node has received Q or more view change messages each indicating that a node that sent one of the Q or more view change messages agrees to the view change message;

multicasting another prepare message indicating an acceptance of the new view message to at least some of the backup nodes including the new primary node; and

obtaining another (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes.

20. The storage medium of claim 18 , wherein, after storing at least the minimal amount of consensus messages and before multicasting the commit message, the operations further comprise:

obtaining, from each node of Q or more of the backup nodes, a view change message indicating that the each node of Q or more of the backup nodes agrees to the view change message;

multicasting to at least some of the backup nodes a new view message indicating that the one backup node acting as a new primary node has received the Q or more view change messages; and

obtaining another (Q−1) or more prepare messages respectively from (Q−1) or more of the backup nodes.