1. 12 Apr, 2008 4 commits
  2. 11 Apr, 2008 3 commits
  3. 10 Apr, 2008 4 commits
  4. 09 Apr, 2008 2 commits
  5. 08 Apr, 2008 3 commits
    • Mike Hibler's avatar
      UT is in production use now. · ac224f78
      Mike Hibler authored
      ac224f78
    • Robert Ricci's avatar
      Revert to revision 1.17 - 1.18, which was rushed for the paper, had a · db3ca65a
      Robert Ricci authored
      bunch of bugs, and wasn't working in production use.
      db3ca65a
    • Leigh B. Stoller's avatar
      A little auth module plugin I wrote to allow cross-domain login from · fc9aae80
      Leigh B. Stoller authored
      emulab to the protogeni wiki, but I use it for all of the wikis. The
      basic idea is that the emulab backend inserts a hash value into the
      cookie table in the trac DB on ops (via ssh). Then Emulab issues a
      redirect over to the trac wiki, with the uid/hash values as arguments
      to the xlogin URL. This hash is use-once; if it exists in the cookies
      table, it is deleted and a new one generated by the underlying auth
      module, and a cookie returned to the browser. The user is thus logged
      in for all subsequent access.
      
      Why? Cause emulab.net cannot insert auth cookies for protogeni.net, so
      must let the auth module inside trac insert the cookie.
      fc9aae80
  6. 07 Apr, 2008 4 commits
  7. 03 Apr, 2008 2 commits
  8. 02 Apr, 2008 5 commits
  9. 01 Apr, 2008 1 commit
  10. 31 Mar, 2008 6 commits
  11. 30 Mar, 2008 3 commits
  12. 28 Mar, 2008 3 commits
    • David Johnson's avatar
      Fixes. · 77878683
      David Johnson authored
      77878683
    • David Johnson's avatar
      Add support to synch with and store plab nodegroups in our DB (plab uses · d0f1cec3
      David Johnson authored
      them to key many different control actions, and we can't ignore them
      anymore).  The driving need in this case is to dist out rootballs based on
      the nodegroup so we can test plab4.2... but other ways to use this will
      likely arise.
      
      Also expanded the notion of config attributes I added during the push to
      support multiple PLCs.  Now, config attributes are just per-PLC or
      per-slice; they are per-PLC, but also conditional across
      slice,nodegroup,node tuples.  This gives a ton of flexibility for the
      different actions we need to take based on what kind of node the node is
      -- and plab typically seems to distinguish this based on nodegroups.
      d0f1cec3
    • David Johnson's avatar
      81a10581