1. 20 Jan, 2009 1 commit
  2. 19 Jan, 2009 1 commit
  3. 16 Jan, 2009 7 commits
  4. 15 Jan, 2009 3 commits
  5. 14 Jan, 2009 2 commits
  6. 13 Jan, 2009 4 commits
  7. 12 Jan, 2009 9 commits
  8. 09 Jan, 2009 3 commits
  9. 08 Jan, 2009 9 commits
  10. 07 Jan, 2009 1 commit
    • Leigh B. Stoller's avatar
      I had such ambitious goals when I started this project! My intent was · cccc74eb
      Leigh B. Stoller authored
      to allow elabinelab vlan firewalls. Alas, I never got there, fell into
      a pit of hell and never made it back out. However, I did do a few
      things.
      
      I did make some changes to deal with stacks, mostly making sure that
      stacks are processed properly inside the elabinelab, and passed
      through to the outer emulab, which currently ignores the stack. See
      the new version of the proxy.
      
      The big change was to -m and -o. These now require a pid/eid argument
      so that we can create (and delete) vlan objects in the DB, for both
      the control and experimental stacks. Most notably, when creating a
      firewalled experiment, we get a Lan(VLan) object for the control
      network fwvlan, and an entry in the softstate vlans table (that
      mirrors what is on the switches). This is intended to make things
      easier to cleanup after a swap error, and to catch inconsitencies
      before we release nodes. Note that -m and -o take -f to override the
      pid/eid requirement, in which case they operate as before.
      
      Trunk enable/disable and Port enable/disable now record that state in
      the new interface_state table.
      
      We now proxy doPortControl().
      
      A second -l option overrides Keiths change that prints only vlans
      associated witl experiments; print *all* vlans.
      cccc74eb