You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
op-devstack: sysgo hard-codes SupportsPostFinalizationELSync=false while running op-reth #23318
Consequence: an ELSync op-node under sysgo behaves as if it drove op-geth. When it starts against an EL that already has a non-genesis finalized block, it takes the geth-only "Skipping EL sync and going straight to CL sync because there is a finalized block" branch instead of re-entering EL sync as it would with op-reth in production (engine_controller.go#L788-L798). Sysgo EL-sync and restart tests therefore exercise a code path production op-reth deployments don't take, and the op-reth path goes untested end to end.
Fix direction: derive the flag from the L2 EL kind sysgo actually starts. For example, map the EL kind to an engine.Kind and use Kind.SupportsPostFinalizationELSync(), or set the op-node engine kind directly so the flag follows as it does in service.go.
🤖 Fable says: Sequencing: this lands as its own PR right after the sysgo op-geth L2 EL is removed (#21744, step E). With op-reth as the only EL kind the flag is simply true for op-node, and a separate PR keeps any change in the sysgo EL-sync tests attributable. Draft #23421 plumbs the same flag through L2CLConfig for the mixed runtime only and touches the same line in singlechain_build.go, so the two need coordinating.
Activity
sebastianst commentedon Nov 12, 2026
🤖 Fable says: Sequencing: this lands as its own PR right after the sysgo op-geth L2 EL is removed (#21744, step E). With op-reth as the only EL kind the flag is simply
truefor op-node, and a separate PR keeps any change in the sysgo EL-sync tests attributable. Draft #23421 plumbs the same flag throughL2CLConfigfor the mixed runtime only and touches the same line insinglechain_build.go, so the two need coordinating.