Have you checked if 1 or more of the worker processes for that topology died?
On Wed, Jun 29, 2016 at 9:09 PM, #ZHANG SHUHAO# <[email protected]> wrote: > Hi all, > > > > > > I’m testing with custom scheduling using Storm v1.0.0, but observed a > weird situation: > > > > At first, my custom scheduler successfully schedule all components of a > topology. > > > > That is “org.apache.storm.scheduler.cluster.needsScheduling(topology);” > returns false. > > > > After a while, it again detect the topology has some components needed to > be scheduled, and re-perform scheduling again. Why? > > > > Thanks. > > Best regards. > > Shuhao Zhang (Tony) > > +65-86166722 > > >
