Context
OpenShift Logging is migrating from Elasticsearch to Loki. While the option to use Loki has existed forquite a while, the information about end of Elasticsearch support has not been available until recently. With the information available now, we can expect more and more customers to migrate and hit the issue described in INSIGHTOCP-1927.
P.S. Note the bar chart in INSIGHTOCP-1927 which shows how frequently is the related KCS linked in customer cases.
Data to gather
LokiStack custom resources (any name, any namespace)
Backports
The option to use Loki is available since Logging 5.5 whose compatibility started at OCP 4.9. Considering the OCP life cycle, backports to up to OCP 4.14 would be nice.
Unknowns
Since Logging 5.7, Logging supports installation of multiple instances in customer namespaces. The Insights Operator would have to look for the CRs in all namespaces, which poses the following questions:

What is the expected number of the LokiStack CRs in a cluster?
Should the Insights operator look for the resource in all namespaces? Is there a way to narrow down the scope?

The CR will contain the name of a customer namespaces which is a sensitive information.
What is the API group of the CR? Is there a risk of LokiStack CRs in customer namespaces that would NOT be related to OpenShift Logging?



SME
Oscar Arribas Arribas

N/A