Is this project compatible on OCP 4.x with RHCOS?
Closed this issue · 10 comments
Describe the bug
I dont know is it a bug or not. So I want to implement koku on my OCP 4.2 cluster which use RHCOS as its worker node. But it get failed:
Aborting, target uses selinux but python bindings (libselinux-python) aren't installed.
I didnt see any possibility to install libselinux-python on RHCOS.
To Reproduce
Steps to reproduce the behavior:
Running this step:
./ocp_usage.sh --setup -e OCP_API="https://api.openshift-prod.mycompany.com" -e OCP_METERING_NAMESPACE="metering" -e OCP_TOKEN_PATH="/path/to/ocp_usage_token" -e METERING_API="https://metering.metering.api.ocp.com"
Failed
Getting this error:
Aborting, target uses selinux but python bindings (libselinux-python) aren't installed.
Hi @riupie . I understand the message may be a little confusing. In this case, those playbooks are referring to localhost as the "target" system. So, the system you're running the korekuta playbooks from needs to have the SELinux bindings installed.
The OCP 4.2 cluster itself should already have the dependencies we require.
Ok, finish installing koku. But then I got this error on my report.
@riupie can you provide output for oc get reportquery?
Also you may want to check out the Cost Management operator. It's now available via OperatorHub from your OCP 4 cluster to make the install process simpler.
Its based on the following project that aimed to make setup of OCP cost metric collection easier: https://github.com/project-koku/korekuta-operator
Output from report query:
[root@ag-koku ~]# oc get reportquery
NAME AGE
cluster-cpu-capacity 22d
cluster-cpu-capacity-raw 22d
cluster-cpu-usage 22d
cluster-cpu-usage-raw 22d
cluster-cpu-utilization 22d
cluster-memory-capacity 22d
cluster-memory-capacity-raw 22d
cluster-memory-usage 22d
cluster-memory-usage-raw 22d
cluster-memory-utilization 22d
cluster-persistentvolumeclaim-request 22d
hccm-openshift-node-labels 7d5h
hccm-openshift-node-labels-lookback 7d5h
hccm-openshift-persistentvolumeclaim 7d5h
hccm-openshift-persistentvolumeclaim-lookback 7d5h
hccm-openshift-usage 7d5h
hccm-openshift-usage-lookback 7d5h
namespace-cpu-request 22d
namespace-cpu-usage 22d
namespace-cpu-utilization 22d
namespace-memory-request 22d
namespace-memory-usage 22d
namespace-memory-utilization 22d
namespace-persistentvolumeclaim-request 22d
namespace-persistentvolumeclaim-usage 22d
node-cpu-allocatable 22d
node-cpu-allocatable-raw 22d
node-cpu-capacity 22d
node-cpu-capacity-raw 22d
node-cpu-utilization 22d
node-labels-raw 7d5h
node-memory-allocatable 22d
node-memory-allocatable-raw 22d
node-memory-capacity 22d
node-memory-capacity-raw 22d
node-memory-utilization 22d
persistentvolume-labels-raw 7d5h
persistentvolumeclaim-capacity 22d
persistentvolumeclaim-capacity-raw 22d
persistentvolumeclaim-labels-raw 7d5h
persistentvolumeclaim-phase-raw 22d
persistentvolumeclaim-request 22d
persistentvolumeclaim-request-raw 22d
persistentvolumeclaim-usage 22d
persistentvolumeclaim-usage-raw 22d
persistentvolumeclaim-usage-with-phase-raw 22d
pod-cpu-limit-raw 7d5h
pod-cpu-request 22d
pod-cpu-request-raw 22d
pod-cpu-usage 22d
pod-cpu-usage-raw 22d
pod-labels-raw 7d5h
pod-memory-limit-raw 7d5h
pod-memory-request 22d
pod-memory-request-raw 22d
pod-memory-usage 22d
pod-memory-usage-raw 22d
pod-persistentvolumeclaim-info-raw 7d5h
[root@ag-koku ~]#
I will check OperatorHub version.
Btw, how to cleanup my koku installation?
You can remove the existing resources with the following command:
oc delete report --selector cost-management=true || true && oc delete reportquery --selector cost-management=true || true && oc delete reportdatasource --selector cost-management=true
i have installed via OperatorHub. Should I create report manually? because it only create report query.
Trying to create report manually and get the same error:
[root@ag-bastion-server koku]# oc get reportquery -n openshift-metering --selector cost-management=true
NAME AGE
cm-node-labels-raw 11m
cm-openshift-node-labels 10m
cm-openshift-node-labels-lookback 10m
cm-openshift-persistentvolumeclaim 10m
cm-openshift-persistentvolumeclaim-lookback 10m
cm-openshift-usage-lookback 10m
@riupie The cost-mgmt-operator should create the reports, no manual intervention should need to have been taken. The following should be created by the operator:
Operator metering is not instantaneous. The reports are hourly, it can take 2+ hours for the dependencies to align and have data.
If you don't have reports, then you should grab the Cost management operator log, in case there is some failure.
If possible, could we close this issue as I believe, we have discussed how korekuta interacts with OCP 4.x (via a RHEL system with Insights that has network connectivity to the cluster). If you continue to have problems with the korekuta-operator can you open the issue there?
Closing so issues regarding the operator can be opened against that repository.
If you have further questions regarding this repository feel free to open another issue.