I've been working on a patch series to overhaul the fscache code in the kclient. I also have this (really old) tracker to add fscache testing to teuthology:     https://tracker.ceph.com/issues/6373 It would be ideal if the clients in such testing had a dedicated filesystem mounted on /var/cache/fscache, so that if it fills up it doesn't take down the rootfs with it. We'll also need to have cachefilesd installed and running in the client hosts. Is it possible to do this in teuthology? How would I approach this?

On Tue, 2020-12-01 at 10:22 -0500, David Galloway wrote: > On 12/1/20 9:55 AM, Jeff Layton wrote: > > I've been working on a patch series to overhaul the fscache code in the > > kclient. I also have this (really old) tracker to add fscache testing to > > teuthology: > > > >     https://tracker.ceph.com/issues/6373 > > > > It would be ideal if the clients in such testing had a dedicated > > filesystem mounted on /var/cache/fscache, so that if it fills up it > > doesn't take down the rootfs with it. We'll also need to have > > cachefilesd installed and running in the client hosts. > > > > Is it possible to do this in teuthology? How would I approach this? > > > > I think I can make this happen pretty easily in ceph-cm-ansible. What > I'd need from you is desired filesystem type and size. Once I'm done > with my end, we'll need to create a cephlab.ansible overrides yaml > fragment to stick in that suite's qa directory. > Ok, cool: fstype: xfs or ext4 (either is fine) size: ~50g or so would be ideal, but we can probably get away with less        if necessary