Unlike a full testnet Merge hard fork, a shadow fork is a smaller test fork that focuses on one or two small changes that need to happen when the Ethereum Merge occurs on the main protocol

Ethereum’s ninth shadow fork went live as the network continues to test its imminent transition from a proof-of-work to a proof-of-stake consensus model.

In preparation for Ethereum’s merge with the proof-of-stake Beacon Chain to complete the transition, it has been undergoing a series of test forks that copy the data from the main network to a test environment network (testnet).

Unlike a full testnet Merge hard fork, like the Sepolia hard fork, a shadow fork is a smaller test fork that focuses on one or two small changes that need to happen when the Ethereum Merge eventually occurs on the main protocol. During this ninth shadow fork, developers focused primarily on testing updates and the releases used in the recent Sepolia hard fork ‘but on a more intensive network,’ said Parithosh, DevOps engineer at the Ethereum Foundation.

Today’s shadow fork occurred at 00:00 UTC when Terminal Total Difficulty (TTD) was overridden at 53945568722258575228928. In the hours since the shadow fork happened, no significant glitches have been reported.

The fork occurred significantly earlier than expected, however. Originally it was predicted to happen at 15:00 UTC, but instead it occurred at 00:00 UTC. This discrepancy was due to a calculator error with the developers’ estimation tool, not because of any error in the code itself.

Parithosh told CoinDesk: There seems to have been a minor spike in hashrate that sped up things a bit. The TTD estimation tool I use is also a local tool, it caches a lot of blocks and tries to average things out, it seems to have a lot of stale state since I haven’t cleared the old state.

He added: There were, however, no problems that were attributed to the TTD being hit earlier. All the nodes were ready except for a few syncing nodes. So this is a good sign that even when TTD is hit much earlier than expected, the network works as expected.

Parithosh also noted that to ensure a similar situation doesn’t occur on mainnet, developers would use multiple TTD estimation methods.