In LHCb, we are using a new HPC in "pre-production" and the connection is quite unstable for now.
To avoid reporting a Done job as Failed just because we cannot get its status or its outputs, we retry to contact the CE a few times.
BEGINRELEASENOTES
*WorkloadManagement
CHANGE: Make RemoteRunner more resilient to CE issues
ENDRELEASENOTES
In LHCb, we are using a new HPC in "pre-production" and the connection is quite unstable for now. To avoid reporting a
Done
job asFailed
just because we cannot get its status or its outputs, we retry to contact the CE a few times.BEGINRELEASENOTES *WorkloadManagement CHANGE: Make RemoteRunner more resilient to CE issues ENDRELEASENOTES