Closed gvbalaji closed 1 year ago
Probably also need packages to represent the actual function instances, including all their requirement CRs, etc., that go through the package specialization process.
/assign @vjayaramrh
Open issues:
@johnbelamaric @henderiw Should the value for the name field at https://github.com/nephio-project/free5gc-packages/blob/main/pkg-example-amf-bp/package-context.yaml#L8 be the name of the package i.e pkg-example-amf-bp
instead of just example
cc: @rravindran123
That gets filled in automatically by kpt/Porch when the package published to a deployment repository (by Porch) or pulled with kpt get --for-deployment
.
Where are the instructions to deploy the free5gc nf packages? I am looking for instructions something similar to step 5 at https://github.com/nephio-project/test-infra/blob/main/e2e/provision/README.md
https://github.com/nephio-project/test-infra/pull/61
I haven't run through and tested all this, but this is quite close to the flow we will have overall.
Update on June 5th:
ubuntu@nephio-r1-e2e:~$ kubectl get pods -A --kubeconfig edge01-kubeconfig
NAMESPACE NAME READY STATUS RESTARTS AGE
config-management-monitoring otel-collector-6cb84fc5d6-79xzr 1/1 Running 0 28m
config-management-system config-management-operator-5c45bf56bb-tc9rl 1/1 Running 0 31m
config-management-system reconciler-manager-86c5f8cc79-nz88t 2/2 Running 0 28m
config-management-system root-reconciler-edge01-bdd54c6f6-kcd7c 4/4 Running 0 28m
free5gc-cp free5gc-ausf-7d494d668d-vbhkp 0/1 Init:0/1 0 17m
free5gc-cp free5gc-nrf-66cc98cfc5-g6lnn 0/1 Init:0/1 0 17m
free5gc-cp free5gc-nssf-668db85d54-9x5qz 0/1 Init:0/1 0 17m
free5gc-cp free5gc-pcf-55d4bfd648-5n6hz 0/1 Init:0/1 0 17m
free5gc-cp free5gc-udm-845db6c9c8-t4cmm 0/1 Init:0/1 0 17m
free5gc-cp free5gc-udr-79466f7f86-7bcx8 0/1 Init:0/1 0 17m
free5gc-cp free5gc-webui-84ff8c456c-45hdv 0/1 Init:0/1 0 17m
free5gc-cp mongodb-0 0/1 Pending 0 17m
kube-system coredns-787d4945fb-x2hjq 1/1 Running 0 31m
kube-system coredns-787d4945fb-xt2jh 1/1 Running 0 31m
kube-system etcd-edge01-5x8xr-sc59z 1/1 Running 0 31m
kube-system kindnet-fd7hg 1/1 Running 0 30m
kube-system kindnet-kmb2d 1/1 Running 0 30m
kube-system kindnet-kvv9c 1/1 Running 0 29m
kube-system kindnet-llk6j 1/1 Running 0 29m
kube-system kube-apiserver-edge01-5x8xr-sc59z 1/1 Running 0 31m
kube-system kube-controller-manager-edge01-5x8xr-sc59z 1/1 Running 0 31m
kube-system kube-proxy-4xm6q 1/1 Running 0 31m
kube-system kube-proxy-5n4tp 1/1 Running 0 29m
kube-system kube-proxy-5q8db 1/1 Running 0 30m
kube-system kube-proxy-nwbvm 1/1 Running 0 29m
kube-system kube-scheduler-edge01-5x8xr-sc59z 1/1 Running 0 31m
resource-group-system resource-group-controller-manager-6cdfc6486d-ldmbd 3/3 Running 0 28m
Need a fix for this to continue further testing to close out the issue. Is there an issue already open for this? If not, I can open one.
@johnbelamaric , Meanwhile, Who can provide the instructions to setup the SMF and UPF so that the ping between them can be tested out. Want to get familar with those instructions until the above pod issues are debugged and resolved?
https://github.com/nephio-project/nephio/issues/256
I think @henderiw and the package specialization team are working on the SMF/UPF/AMF integration.
Yes open PR(s) Are:
https://github.com/nephio-project/api/pull/38 -> this allows to validate empty as not ready. Right now I use empty string.
https://github.com/nephio-project/api/pull/39 -> this allows for ipv6 and ipv4. So this allows us to specify what we want rather than hardcoding. It is a small change I want to incorporate after #259 is approved
so I have the specialization running, but to continue the PR chain I need the above 3 approved to proceed.
| kind: UPFDeployment | metadata: # kpt-merge: /upf- | name: "upf-cluster03" | annotations: | internal.kpt.dev/upstream-identifier: 'workload.nephio.org|UPFDeployment|default|upf-' | spec: | capacity: | maxDownlinkThroughput: 10G | maxUplinkThroughput: 10G | interfaces: | - name: n3 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | - name: n4 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | - name: n6 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | networkInstances: | - name: vpc-internal | interfaces: | - n4 | - name: vpc-internet | dataNetworks: | - name: vpc-internet | pool: | - prefix: 172.0.0.0/8 | interfaces: | - n6 | - name: vpc-ran | interfaces: | - n3
| kind: UPFDeployment | metadata: # kpt-merge: /upf- | name: "upf-cluster03" | annotations: | internal.kpt.dev/upstream-identifier: 'workload.nephio.org|UPFDeployment|default|upf-' | spec: | capacity: | maxDownlinkThroughput: 10G | maxUplinkThroughput: 10G | interfaces: | - name: n3 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | - name: n4 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | - name: n6 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | networkInstances: | - name: vpc-internal | interfaces: | - n4 | - name: vpc-internet | dataNetworks: | - name: vpc-internet | pool: | - prefix: 172.0.0.0/8 | interfaces: | - n6 | - name: vpc-ran | interfaces: | - n3
| kind: UPFDeployment | metadata: # kpt-merge: /upf- | name: "upf-cluster03" | annotations: | internal.kpt.dev/upstream-identifier: 'workload.nephio.org|UPFDeployment|default|upf-' | spec: | capacity: | maxDownlinkThroughput: 10G | maxUplinkThroughput: 10G | interfaces: | - name: n3 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | - name: n4 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | - name: n6 | ipv4: | address: 10.0.0.10/24 | gateway: 10.0.0.1 | vlanID: 10 | networkInstances: | - name: vpc-internal | interfaces: | - n4 | - name: vpc-internet | dataNetworks: | - name: vpc-internet | pool: | - prefix: 172.0.0.0/8 | interfaces: | - n6 | - name: vpc-ran | interfaces: | - n3
So the steps for me are:
I had to do a little hack for now based on https://github.com/nephio-project/nephio-example-packages/blob/main/cluster-capi-kind-docker-templates/docker_machine_template_worker.yaml#L12 that is mountable to make progress.
I had to create a directory /opt/cni/bin/edge01/mongodb and modify the PV yaml file as below to point to the newly created directory and apply the yaml file to create a PV for now
ubuntu@nephio-r1-e2e:~$ cat pv-edge01.yaml
---
apiVersion: v1
kind: PersistentVolume
metadata:
name: pv-loc-sc
spec:
persistentVolumeReclaimPolicy: Delete
#storageClassName: local-path
capacity:
storage: 7Gi
accessModes:
- ReadWriteOnce
local:
path: "/opt/cni/bin/edge01/mongodb"
nodeAffinity:
required:
nodeSelectorTerms:
- matchExpressions:
- key: kubernetes.io/os
operator: In
values:
- linux
June 6th Update: Followed instructions at https://github.com/nephio-project/test-infra/blob/main/e2e/provision/README.md and was able to execute steps till step 9 with workarounds. Step 10 that attempts to deploys UPF does not deploy any pods. Need instructions to deploy SMF as well
Depends on WIm's hydration work and operator work from Simon merged
So, we have free5gc-cp up and running. We have two more PRs to merge to get the operator up and running. And I think Wim nearly has the hydration/specialization work done.
Sequence of operations :
This is working now. There is still some flakiness, I think due to concurrency issues. But overall it does work.
Make sure all PODs are up and running Make sure ping works between SMF and UPF