Something I was thinking about that could be used in either case, is a latency monitoring algorithm similar to SED (Sandstorm) that Adam and I were discussing a while ago.
Basically, the job scheduler monitors the amount of time it takes to run jobs. Once the average time exceeds a threshold, jobs are postponed for a while, then tries them again. So, the scheduler would postpone jobs on a server that's under heavy load. -Adrian --- On Sun, 10/17/10, Scott Gray <[email protected]> wrote: > Hi All, > > I was looking at the Quartz Scheduler project > (www.quartz-scheduler.org) over the weekend and it looks > like it could be a good fit for OFBiz. We ran into > some issues with the scheduled service code in OFBiz > recently where a heavy server load would cause all sorts of > strangeness (multiple reschedules for a single failed job, > inability to purge old jobs before a timeout, those two > combined eventually bringing the server to its knees), and > my options are to either find and fix the problem(s) or > replace the scheduler with an external solution. > > I've only had a brief look but it appears like quartz is > pretty extensible and would allow us to continue to support > things like temporal expressions (and the deprecated > recurrence infos) and could probably increase the number of > scheduling features available to OFBiz. It's ASL2 > licensed and seems to be pretty mature. > > Does anyone have any experience with quartz to share? > Opinions or other possible alternatives would be most > welcome. I'm not looking to implement anything anytime > soon but figured we may as well start discussing it. > > Thanks > Scott > > HotWax Media > http://www.hotwaxmedia.com > >
