Around https://github.com/Kantigen/ka-server/blob/master/lib/KA/App/MQWorker/Daemon.pm#L88 we try to run a job, and, if an exception is thrown, we bury the job and continue.
At the very least, the failed job should be logged: the exception, call trace (if possible), and the job payload. And anything else that may come in handy (number of timeouts, reserves, kicks, etc.).
Better yet would be to somehow email the admins, whether in-game or not, with the details, so that we know to look for it.
In theory, this should never happen - but that makes it all that much more important to throw it up in our faces if it does happen. We may not notice otherwise.
Around https://github.com/Kantigen/ka-server/blob/master/lib/KA/App/MQWorker/Daemon.pm#L88 we try to run a job, and, if an exception is thrown, we bury the job and continue.
At the very least, the failed job should be logged: the exception, call trace (if possible), and the job payload. And anything else that may come in handy (number of timeouts, reserves, kicks, etc.).
Better yet would be to somehow email the admins, whether in-game or not, with the details, so that we know to look for it.
In theory, this should never happen - but that makes it all that much more important to throw it up in our faces if it does happen. We may not notice otherwise.