Description of problem:

BF-2 has 16G mmc, which can be an issue when we run for a long time (logs fill the disc)

Version-Release number of selected component (if applicable):

4.11,4.12

How reproducible:

When BF-2 cluster running for a long time 

• Document to configure policy so that "1 out of 16" vf issue doesn't happen
• Document reboot of 2-cluster design
• Emphasize that the Infra cluster OCP should use the DPUs uplink network as the primary network (and not the OOB).
• Document that DPU needs to be set to ethernet mode (to ensure it's not in IB mode)
• Document adding pci=realloc
• DPU docs need to mention log rotation