When GlusterFS detects a serious problem with a brick, it logs a message and the output of gluster volume status changes. It would be awfully handy to be able to configure a hook script that is fired on brick failure (or perhaps other issues) so administrators are alerted immediately. Such a script might fire an SNMP trap or send an e-mail to the operations team, and it would be up to the customer to implement it as appropriate in their environment. The current alternative (log parsing, gluster volume status parsing) is comparatively error-prone.
Feature requests make most sense against the 'mainline' release, there is no ETA for an implementation and requests might get forgotten when filed against a particular version.
Migrated to github: https://github.com/gluster/glusterfs/issues/588 Please follow the github issue for further updates on this bug.