Skip to content

20180502_keynote_3

20180502_keynote_3

k8s @CERN

Motication for Federation
* Periodic load spike
* Campaign of data reconstruction before conferences …

Use case1

Tools used
* HTCondor
* Sched, collector, negotiator, StartD (Running workload in node)

Config Example:

CERNEnvironment=
Datacenter=
TotalCPUs=
TotalMemoru=

K8sHost with HTCondor control plane
* Sched
* Collector
* Negotiaor

kubefed init fed --host-cluster-context=

kubefed join --context= \
             --host-cluster= \
             --cluster-context=

StartD is deployed as a Daemonset

Use case2

OpenSouce tools for Reusable Analysis Platform
* github.com/recast-hep: Live output
* github.com/diana-hep/yadage: workflow orchestrator
* github.com/reanahub: Registry to share results