Skip to content

[Upstream nexum-runtime] Chain-log reconnect misses reorgs landing in the disconnect window #647

Description

@lgahdl

Tracking issue — the bug and its fix live upstream in nullislabs/nexum-runtime, which shepherd-engine depends on as a pinned git dependency.

Upstream issue: nullislabs/nexum-runtime#62

Found during an internal red-team review (malicious/compromised RPC provider persona): a chain-log poller reconnect resumes just past the last delivered block instead of re-validating it, so a reorg landing exactly in a disconnect window is never surfaced as a removed event. Any shepherd module consuming chain-log subscriptions (e.g. the composable-cow keeper) inherits this gap.

This issue tracks: re-verifying the fix once nexum-runtime ships it, and bumping shepherd's pinned dependency revision.

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working.component/chainalloy provider, chain host backend, RPC middleware

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions