Describe the bug
For some reason leader Rotation is choosing the same bls key despite the fact it already failed election
To Reproduce
Steps to reproduce the behavior:
- Check the logs of the internal validator in Loki or on the node
- Notice that on the gap:1 and gap:2 is the same bls key:
gap: 1:
"lastLeaderPubKey": "356ea6140756c46ce6ff89c49a831a0c72e6ba6a8577ebd63918426f67a7def02ab0c5cfcc740df3293b62a5e6980094",
"leaderPubKey": "d4e50a7226b2ee0f921ebb52ab3fa5f2a9d44046c4f6ec29183c082ff1b822914ef48b597b00b32df38a784f58552018",
"gap": 1
gap: 2:
"lastLeaderPubKey": "356ea6140756c46ce6ff89c49a831a0c72e6ba6a8577ebd63918426f67a7def02ab0c5cfcc740df3293b62a5e6980094",
"leaderPubKey": "d4e50a7226b2ee0f921ebb52ab3fa5f2a9d44046c4f6ec29183c082ff1b822914ef48b597b00b32df38a784f58552018",
"gap": 2,
- Please, note that SenderPubKey was the 50b20b4eb89041aa4715e87e4aa6430e674658959d8be657d43e4d3d4c35d7507d6a40aec5188218dcbb5460e354fd12 in both cases
Expected behavior
A new bls key should be proposed on the the next election
Environment (please complete the following information):
- OS: Linux
- Go environment [
1.24.2 ]
Additional context
Add any other context about the problem here