We could automatically recovery from checksum failures encountered during WAL recovery by filtering out any subsequent entries for any WriterId that has failed a check.
This will of course threaten consensus as log loss may result in a member being elected that causes committed entries to be thrown away so perhaps should either be a configuration setting that is enabled after encountering an error.
We could automatically recovery from checksum failures encountered during WAL recovery by filtering out any subsequent entries for any WriterId that has failed a check.
This will of course threaten consensus as log loss may result in a member being elected that causes committed entries to be thrown away so perhaps should either be a configuration setting that is enabled after encountering an error.