Daniel Gryniewicz

Most or all of those functions are doxygen documented in fsal/commonlib.c. Basically all of our API-level documentation is done via doxygen comments on the function's implementation.

Unfortunately, no. The semantics of the NFS protocol require that we be able to open by handle, and the linux kernel devs have decided that those APIs need a CAP...

internet seems to think "malloc(): unaligned tcache chunk detected" is memory corruption.

ceph_init() failed. This appears to set up basic communication with the cluster (connecting to the monitors, getting the MonMap, etc.) You will need to check the ceph logs to find...

This is already submitted here: https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/1194724

Ganesha doesn't support WRITE_SAME. If you try it, you'll get NFS4ERR_NOTSUPP back.

ENOTCONN means the socket isn't connected to the gluster cluster? This seems unlikely, since it shouldn't get far enough to try a READDIR without a connection. FSAL_GLUSTER never creates a...

Unfortunately, Red Hat stopped supporting Gluster years ago, and Red Hat was the only company doing Ganesha on Gluster work, so there's no one in the community that's supporting FSAL_GLUSTER...

FSAL_CEPH has support for ACLs, but it's conditionally compiled. Maybe your build has it turned off?

-61 is ENODATA. It looks like CephFS returns that when the requested xattr (either "system.posix_acl_access" or "system.posix_acl_default" in this case) doesn't exist. I'm not sure why it wouldn't exist, or...