Ethereum’s second mainnet shadow fork passed its Merge rehearsal

Ethereum developers completed the second mainnet shadow fork’s transition to proof-of-stake on April 23, 2022, producing an important rehearsal for the network’s planned Merge without changing Ethereum’s canonical mainnet.

The test copied mainnet state and transaction history into an isolated environment, then directed participating nodes to follow the proposed post-Merge rules. Developer Adrian Sutton reported on April 23 that the transition itself completed successfully despite client problems encountered while processing deposits. The result did not establish that the production Merge was ready, but it gave client teams a more realistic test than a purpose-built network with limited history and activity.

What the shadow fork tested

Ethereum’s planned Merge required two previously separate systems to operate together: the execution layer, which processed transactions and smart-contract state, and the Beacon Chain’s proof-of-stake consensus layer. A shadow fork let developers test that handoff using conditions derived from the live chain while containing failures inside a separate network.

The preserved configuration repository identifies the environment as `mainnet-shadow-fork-2`. Unlike a production hard fork, the exercise did not ask ordinary Ethereum users, exchanges or application operators to migrate, and proof-of-work miners continued producing canonical mainnet blocks. Claims that Ethereum itself switched consensus mechanisms on April 23 would therefore be incorrect.

The institutional importance was operational. Ethereum depended on independently developed execution and consensus clients, so a credible rehearsal had to test combinations rather than a single reference implementation. A failure limited to one client could still disrupt validators, block production or finality if that software held meaningful network share.

The result and the defects

Sutton’s contemporaneous account said the actual transition proceeded despite late blocks and deposit-processing problems. Prysm produced blocks late enough to cause missed head votes and weak sync-committee participation, while Nimbus produced invalid blocks. Sutton characterized both as deposit-processing issues rather than failures of the Merge mechanism and reported that both had been fixed.

After those fixes, developers observed high participation in attestations and sync committees. Contemporary technical reporting also described the exercise as the first mainnet shadow fork in which every tested client combination survived the transition and remained synchronized afterward. That was a significant compatibility result, but it was not equivalent to a security audit or a guarantee that the same software would behave identically under unrestricted production load.

The shadow fork also exercised nodes whose execution layers were still synchronizing. That created a practical test of optimistic synchronization and of the safety boundary between consensus-layer progress and execution-layer verification—conditions that a cleaner laboratory network might not reproduce.

Why it mattered on April 23

By April 23, repeated Merge delays had made working multi-client evidence more consequential than another projected activation date. The second shadow fork reduced one category of implementation uncertainty and exposed defects while recovery remained inexpensive. It also supplied evidence for subsequent core-developer decisions about additional shadow forks, public testnet transitions and the proof-of-work difficulty schedule.

No authoritative event-day dataset reviewed for this reconstruction establishes that the test caused a measurable ether price move. Crypto trades continuously across venues, and any such claim would require a named trading pair, venue or index, UTC measurement window and causal limitations. This reconstruction therefore makes no price or return attribution.

Later context

Later records show that testing continued through additional shadow forks and public testnets before Ethereum’s canonical mainnet completed the Merge on September 15, 2022. That later outcome confirms the place of the April 23 rehearsal in the testing sequence; it does not convert the April result into a contemporaneous guarantee of production readiness.