At 04:09 AM 8/3/2008, Amos Shapira wrote: >What's defined as "sufficient access"? Is a stat(2) on the file enough >to cause the data push or does it actaully have to be open(2)'ed or >does it even actually have to be completely read(2)? > >Maybe a simple "find ... -size..." on the shared tree during boot >could trigger a full sync? http://www.gluster.org/docs/index.php/Understanding_AFR_Translator look under "Preemptive Self-Heal" for a couple examples.