In production it seems that dying workers (due to exceptions) are not always restarted. In some cases there are no 'restarting' messages at all in nohup.out despite most workers having disappeared.
Production is running node 0.8.2 and latest node_modules as of today.
Version: unspecified
Severity: normal