Version was:
hyperledger/besu 22.7.RC1-SNAPSHOT-openjdk-11-debug 3cae566a8e63 38 hours ago
And they had been up for 26 hours.
So far nothing more than some normal behavior we could maybe make look better in the logs. Then I started checking the proposals coming from besu, and they are empty. All the CLs connected to besus are proposing empty blocks. They should be full moments after TTD on mainnet, plenty of txs to use.
About 90 minutes after TTD, I noticed that finalization times were really unstable. about 15 minutes after that, we just stopped finalizing. Thats when the panic set in, and I started checking every besu node in the mnsf7 group. Timestamps and hashes for each failure are below.
I also tried bouncing both the besu and teku on besu-teku2 just for fun, I did not expect anything to happen since besu is now halted on the block it thinks is bad.
Somehow, the network managed to finalize again later in the evening around 9:34 UTC. Finalization times were still unstable for a couple of epochs, and then stopped again.
at 10:55 UTC I tried bouncing just the teku again to see if it was related to the phantom recurrence of finalization. It had no impact on the network, as expected.
And now for the failures: