-
Notifications
You must be signed in to change notification settings - Fork 16
CVMFS installation on K8s clusters
Fernando Barreiro edited this page Oct 22, 2020
·
3 revisions
Currently the most stable solution is the PRP driver https://github.com/PanDAWMS/prp-osg-cvmfs.
- For GKE some configurations are adapted under the gke branch.
- For other clusters tried so far the configuration in the master branch is desired.
Follow the instructions on the README.txt
git clone https://github.com/PanDAWMS/prp-osg-cvmfs.git
cd prp-osg-cvmfs
git checkout gke
cd prp-osg-cvmfs/k8s/cvmfs
cat README.txt
- Run the test job at the end of the instructions to see CVMFS gets mounted on a pod. This test job simply sleeps so you can log in and look around a bit.
kubectl create -f test/test-job-cvmfs.yaml
kubectl get pods
NAME READY STATUS RESTARTS AGE
grid-job-wwrvz 0/1 ContainerCreating 0 6s
...
[root@aipanda169 test]# kubectl get pods
NAME READY STATUS RESTARTS AGE
grid-job-wwrvz 1/1 Running 0 115s
[root@aipanda169 test]# kubectl exec -it grid-job-wwrvz /bin/bash
[root@grid-job-wwrvz /]# cd /cvmfs/
atlas-condb.cern.ch/ atlas-nightlies.cern.ch/ atlas.cern.ch/ grid.cern.ch/ sft.cern.ch/
[root@grid-job-wwrvz /]# cd /cvmfs/atlas.cern.ch/
.cvmfsdirtab repo/
[root@grid-job-wwrvz /]# cd /cvmfs/atlas.cern.ch/repo/
.images/ benchmarks/ containers/ sw/ tools/
ATLASLocalRootBase/ conditions/ dev/ test tutorials/
Getting started |
---|
Installation and configuration |
Testing and running |
Debugging |
Work with Middleware |
Admin FAQ |
Development guides |
---|
Development workflow |
Tagging |
Production & commissioning |
---|
Scale up submission |
Condor experiences |
Commissioning on the grid |
Production servers |
Service monitoring |
Auto Queue Configuration with CRIC |
SSH+RPC middleware setup |
Kubernetes section |
---|
Kubernetes setup |
X509 credentials |
AWS setup |
GKE setup |
CERN setup |
CVMFS installation |
Generic service accounts |
Advanced payloads |
---|
Horovod integration |