We are upgrading our NFS v3 clients to Scientific Linux 7, (pretty much the same as Centos 7) and running into the problem that running Gnome or Mate cause the Truenas server to fail, with a large number of "lockd not responding" messages. If we kill the client process, then the Truenas recovers after a few minutes. This did not happen with any previous Linux client. Our first workaround was to turn off locking on home directories, but eventually we figured out that specifying XDG_CONFIG_HOME in ~/.config/user-dirs.dirs to point to a file on a mount that did not support locking was sufficient. Still we can't help but think this is likely due to some other configuration error on our part and that someone here might have a suggestion.
I am posting here rather than contacting iXsystems support for the Truenas because we don't want to kill the system, even for a few minutes, to reproduce the problem for iX now that we have a workaround.
Daniel Feenberg
NBER
I am posting here rather than contacting iXsystems support for the Truenas because we don't want to kill the system, even for a few minutes, to reproduce the problem for iX now that we have a workaround.
Daniel Feenberg
NBER