AD in large deployments fills up /var/tmp/.cache quickly; creates incomplete user cache

[gcooper]

The cache available for LDAP + AD + etc is limited to 1.4GB today and if you do a wbinfo -u lookup on a large tree, /var/tmp/.cache will fill up, spewing "filesystem is full" errors on the screen and various commands like wbinfo -u, etc will fail.

Prereqs:

• Install FreeNAS 8.0.2-p1.
• Create an AD forest with over 1000 users.
• Configure AD, but don't start it.

Procedure:

• Execute the following commands on the CLI:

  sh
  umount /var/tmp/.cache
  m=`mdconfig -a -t malloc -s 10m`
  newfs $m
  mount /dev/$m /var/tmp/.cache
  

• Restart the activedirectory service via the GUI.
• Run 'wbinfo' or 'getent passwd' on the CLI.

Expected result:

• The cache should be filled as much as possible and if not possible the user should be alerted to a caching issue, and if this fails it should fall back to streaming the user information from the AD DC as much as permitted within memory constraints on the NAS box and policy on the AD DC server.

Actual result:

• "/var/tmp/.cache: filesystem is full messages" end up on the console.
• Incomplete cached data is obtained and presented on the GUI.

[jhixson]

If you need more cache, it is tunable, that was by design. Currently this cannot be done from the UI but that will eventually change.