Hi all, is there a way to wait for resources to be...
# random
a
Hi all, is there a way to wait for resources to become available on job restart while using default scheduling and standalone mode? Problem: if one of the TMs is unable to cancel task within cancellation timeout during job failover - this TM will be killed. This leads to the situation when job initialization fails with
NoResourcesAvailableException
if new TM didn’t fully launched yet.
slot.request.timeout
has no effect here. Flip for declarative resource management mentions waiting for resource to arrive one of next steps, but I was not able to find changes in this area.