Ceph » CephFS

Creating a ticket for this ongoing work in the interests of clarity.

The idea is that when we get an error in fetch() (or equally detected by forward scrub), we will want to remember that so that:

• we don't go trying to fetch it again and again
• we can quickly brush of clients with an EIO when they go near it
• we can show the list of damage to the user so that they can potentially go repair it and/or have this list be consumed by an auto-repair tool.