Service which collects Platform services and Flow steps logs. It stores them in the Elasticsearch (with GrayLog as an intermediate).
Not critical
The downtime means no logs from platform pods and flow pods. In most cases it is acceptable to stop for several minutes.
Should be started at all VMs used to run the Platform (including VMs for flows and the Platform services).
Use rolling release. Remove the pod and create a new one.
The service would not start without the Kubernetes API service.
Fluentd service would be degraded if the GrayLog is down or unreachable.