The Kintsugi network, on which the bug occurred, was launched less than a month ago.
Preparations for the launch of Ethereum 2.0 are underway on this testnet.
One bug caused Ethereum 2.0 testnet Kintsugi to fork into at least 3 different networks. This causes that network to take 3 days without being able to validate blocks correctly.
The downside was due to a bug-check test within the network, the tool of which is known as Fluzzer. This aims to create invalid blocks by changing certain characteristics, in order to verify that the validators can identify the failure and invalidate the block. This, as detailed by Ethereum developer Marius van der Wijden, in a thread from Twitter.
However, it seems that some of these validators, which were running clients like Nethermind and Besu, identified invalid blocks as correct, creating a new chain, as the Geth client validators did reject them.
But the network partition was much worse, because the validators running Teku, another of Kintsugi’s clients, also marked as valid bad blocks, creating a new fork.
The programmer commented this August 7 through his Twitter, that the creation of invalid blocks was maintained for 2 days, and that, for the moment, the chain is still had errors, which kept developers working on finding them.
At the time of writing, developer van der Wijden has not given updates on whether the network is no longer forked. For January 8 he shared a thread from Twitter on the status of the problem, in which they commented that they had identified what happened.
Regarding Kintsugi, it is worth mentioning that this testnet was launched last December, as reported by CriptoNoticias. In this network the first instances of what will be Ethereum 2.0 are put to the test, with proof of participation (PoS).
Ethereum 2.0 mainnet forked, is it possible?
The error in question that produced this large fork was intentionally caused, however, The results were not as expected. Although it was deliberately produced, such a scenario could occur on the Ethereum 2.0 mainnet.
According pick up the CryptosLate portal in an interview with Marius van der Wijden, the programmer commented that such an error would be “very annoying” on the main network. Although they are detectable errors, the levels of forking caused, could end in the total failure of the network.
However, for the programmer, these types of scenarios that occurred in the testnet are a good sign since it shows the response times of the developers to errors. In addition, it highlights that situations like this do not affect or delay the arrival of Ethereum 2.0.