LinstorCluster clean up and writes to node /var folders

LINBIT SDS Integrations
,
1

When running up piraeus operator I got a list of errors by running the linstor error-reports command on the linstor controller. After completely removing the LinstorCluster and the piraeus operator and then redeploying, I found that the old error reports were still there. On investigation, can see that the error logs are actually getting placed on the nodes under /var/log/linstor-satellite folder. I dont know if this is an issue with my install or by design. If the later this is unwanted configuration? On OpenShift the only processes directly writing ie to /var/log/XXX should be platform workloads, not third party operators.

My point is as much about managing the lifecycle of the piraeus operator as it might be configuration of operator on the platform. Clean deleting (and restoring) of the operator does not seem possible at least in the community edition, many linstor and piraeus api type resources need to be manually cleaned up to re-deploy clean-slate.

UPDATE - answer my own question. Found that rook-ceph writes to host var path? Looks like an explicit setting:

```
apiVersion: ceph.rook.io/v1
kind: CephCluster
metadata:
name: rook-ceph
namespace: rook-ceph
spec:
cephVersion:
image: Quay
dataDirHostPath: /var/lib/rook # Host path where metadata/config is saved
…
```

2

On OpenShift the only processes directly writing ie to /var/log/XXX should be platform workloads, not third party operators

Noted. We chose to use /var/log/… to have persistent errors and logs for this component. Otherwise, if there is a problem with the satellite, and the Pod gets recreated, you lose all previous logs, making it hard to pin down what went wrong.

If this is undesired, you can change the satellite pod template so the var-log volume is an emptyDir type.

My point is as much about managing the lifecycle of the piraeus operator as it might be configuration of operator on the platform. Clean deleting (and restoring) of the operator does not seem possible at least in the community edition, many linstor and piraeus api type resources need to be manually cleaned up to re-deploy clean-slate.

Yes, this is a known gap in the documentation. Not automatically removing everything is a choice to ensure people do not accidentally delete their data, but that means you need additional instructions to “clean up everything”. And those are missing.

We will try to improve this situation in the future.