1. Proposed title of this feature request

OpenShift Logging 6.x enhance Vector profile for log reliability

2. What is the nature and description of the request?

We want to ask for extending the possibility on configurations when running the deliveryMode: AtLeastOnce as follows:

• adding max_size to ensure once the output destination is temporarily not available, to keep on collecting logs
• adding message_timeout_ms=0 which should ensure, messages are not rotated out of the disk buffered not send logs

With these two configuration options we have been able to verify that:

• disk buffer files are created and rotate on the default 128M size until the output destination is available again
• messages are not rotated out when the 128M size is reached 
• messages have been delivered successful after the output destination was brought up online

Compliance to a set of Banking and Financial rules.

3. Why does the customer need this? (List the business requirements here)

Compliance to a set of Banking and Financial rules.

4. List any affected packages or components.

OpenShift Logging Operator 6.x

5. Executed tests at Customer site

Our tests showed that:

•  The disk buffers do not seem to impact the nodes even if reaching the max_size as that is the point where logs will be rejected by the collector
• The disk buffers shall not exceed the OCP node disk pressure conditions in the sense of not filling all ephemeral storage (documentation hint)
• Flushing the buffers was not impacting the node and new logs have been collected again after freeing the block condition on the collector