While there have been NFS gateways over CephFS for a long time, scaling that service across multiple nodes has always been a challenge. Recently, a new recovery backend was merged into the nfs-ganesha userland NFS server that allows a cluster of NFS servers to coordinate their recovery periods using a shared RADOS object, allowing us to scale out a cluster of NFS servers in a loosely-coupled fashion on top of CephFS.
This presentation will cover some basics about NFS recovery, how we solved the problem of coordinating the recovery across a cluster of NFS servers, and some practical deployment scenarios.
Jeff Layton is a longtime Linux kernel and Ceph developer who specializes in network filesystems. He maintains the Linux kernel CephFS client and is a contributor to many other open source projects.
Saturday January 26, 2019 5:00pm - 5:25pm CET
G202
