"Rescuing" a workflow invocation?

When I update tools I occasionally have a failed tool run due to e.g. tool whose Singularity container build has not succeeded yet. Is there any way to re-execute the tool to ensure that the workflow invocation ends up being successful? I’d prefer to not lose the whole workflow run. (This is on a Galaxy 21.01 server but I’m interested if this can be done with a later version of Galaxy)