On 20 May 2019, at 16:51, J. Bruce Fields wrote:
On Mon, May 20, 2019 at 10:22:00AM -0400, Benjamin Coddington wrote:
Ok, I just noticed that we set fl_owner to the nlm_host in
nlm4svc_retrieve_args, so things are not as dire as I thought. What
would be nice is a sane set of tests for NLM..
What would we have needed to catch this? Sounds like it turns
multi-client testing wouldn't have been required? (Not that that
would
be a bad idea.)
Two NLM clients would be ideal to exercise the full range of expected
lock behavior. I suspect that's something I can do with what's in pynfs
today, but I haven't looked yet. I suppose if there's a test for NLM I
should make one for v4 too..
Ben
