I am running into continual problems with Gearman and TheSchwartz when trying to upgrade my 2005 vintage livejournal site. I have a small site so everything is running on the same Red Hat box. The latest problem is that some of the workers (load-friends-gm, taglib-gm) die with no diagnostics printed as to why. The documentation is very sketchy on this part... What is the recommended practice for starting the workers: A script that just starts them all then exits, or a script that monitors them to ensure they stay running, or run them from init? If a worker dies it seems to bring down the whole site... what does livejournal.com do to mitigate this? Have they proven very reliable there? Is there any more documentation than the LiveJournal manual? Thanks.