Because we don't care about the child dying in a way that kills redgrid itself, we should handle timeouts as something that kills the redo process, starts a new one, and keeps retrying, rather than killing the redgrid process itself and possibly taking down the node.
Because we don't care about the child dying in a way that kills redgrid itself, we should handle timeouts as something that kills the redo process, starts a new one, and keeps retrying, rather than killing the redgrid process itself and possibly taking down the node.