Merge Train Pipeline - Can't retry failed job

I have a job that failed in a merge train pipeline due to an RPC error (pictured below). It seems that I'm not able to retry the single job that failed. Normally when this happens I just retry the job and it succeeds. Is this possible with a merge train pipeline?

Maybe the RPC error is what needs to be fixed? I've tried to reduce the Git shallow clone to 10. We do run our own runners using AWS spot instances could this be the issue?