Neo: Abnormal block time per 7 height

[i359]

[i359]

Abnormal block time per 7 height，similar problems have occurred many times before，what‘s happen？

[Tommo-L]

Thanks, we're also waiting for the consensus node to restart.

[shargon]

One cn it's down

[i359]

How to avoid similar things caused by node failure that greatly extend the block time? thanks！

1. How to monitor the node status and repair in time?

2. If there is a fault node, can it be identified and skipped?

[i359]

Abnormal block time per 7 height as below，thanks！

[EdgeDLT]

This is expected behaviour of dBFT when one of the consensus nodes has failed. There is an exponential backoff on block times to ensure liveness when a node fails to get a block produced and signed within the time limit for a consensus round. It was the failed node's turn to produce a block, it did not, so the next node had to produce it (in the next round, i.e. doubling the block time).

I recommend reading this paper prerelease, it explains the inner workings and history of dBFT, then goes into current research directions for a "dBFT 3.0" which aims to improve on several aspects, including block time consistency.

[i359]

Maybe one node down some days，need to repair，thanks！

[i359]

Abnormal block time:

[i359]

Abnormal block time，what's happen？

[Tommo-L]

There have been some attacks, someone send a lot of empty txs.

[i359]

TPS < 2, is it attack?

[Tommo-L]

And also, some of the consensus nodes have timeouts. But they have already been restarted now.

[i359]

Hope to see that each block has 1000 + data, and the block is still stable, thanks！

[i359]

Abnormal block time，what's happen？

[i359]

Abnormal block time per 7 height

[vncoelho]

We plan to solve that with dBFT 3.0.

https://www.mdpi.com/1999-5903/12/8/129
If you have some time to take a look at this paper.
We are polishing the description with some more precise specification for @neo-project/core to move forward with it.

[i359]

@vncoelho ，It is ok for long term solution，but now maybe 1 node down some days，need to fix，thanks！

[Tommo-L]

but now maybe 1 node down some days，need to fix

Yes, it'll be restored in next week, one cn node need to update to the latest version.

[i359]

Maybe 1 node down，need to fix，thanks！

[ProDog]

Can you describe how you do it? the time shown here is normal:
https://neotube.io/block/6130882/page/1
https://neotube.io/block/6130875/page/1

[i359]

Blocktime show as below for your reference, thanks!

https://neoscan.io/blocks/1

[ProDog]

This may be a bug of neoscan.

[i359]

No, neoscan is OK

[ProDog]

This is normal😅:
https://dora.coz.io/blocks

[i359]

@Tommo-L ,@shargon ，Maybe 1 node down some days，need to fix, thanks！

[Tommo-L]

Yes, one consensus node lost. 😂 We need to vote for a new consensus node.

[i359]

Maybe one consensus node lost，pls dev check and fix， thanks！