Running operators in their own namespace
Opened this issue · 2 comments
Currently operators (OpenShift Pipelines for now) are installed in openshift-operators namespace.
There are benefits of installing the operators in their own namespaces (still separated from operands) rather than openshift-operators. It would however require an OperatorGroup first to be created.
- better segregation
- dedicated InstallPlan, which provides for instance the possibility to have different upgrade strategy by operator. This is really useful when one runs into troubles.
- operartors with different scopes don't cohabit very well in the same namespace
It is not an issue as long as we don't install other operators but things may change.
It is not an issue as long as we don't install other operators but things may change.
There's a good chance that we will install other operators in our base deployment on workload clusters. The first to come to mind are the ElasticSearch operator and the OpenShift Logging operator: https://docs.openshift.com/container-platform/4.10/logging/cluster-logging.html#cluster-logging-about_cluster-logging.
These come to mind as our "first option" for forwarding TaskRun logs.
We will be fine with these ones as they get installed by default under the following namespaces:
- openshift-operators-redhat
- openshift-logging