In some cases requests to get blame will call out to the SCM to retrieve it. This happen if storing blame is disabled globally (via the fisheye.blame.calc.enable system property), or for SVN if the changeset is not completely indexed yet.

The fetched blame is used by that request, but not persisted anywhere, so subsequent calls will reach out to the SCM again, until the blame is available from indexing (or always in case storing blame is completely disabled).

The retrieved blame should be persisted instead, so that subsequent calls don't need to re-fetch it.