1. 07 Mar, 2008 2 commits
  2. 06 Mar, 2008 1 commit
  3. 05 Mar, 2008 1 commit
  4. 03 Mar, 2008 3 commits
  5. 29 Feb, 2008 3 commits
  6. 28 Feb, 2008 1 commit
  7. 27 Feb, 2008 3 commits
  8. 26 Feb, 2008 2 commits
    • Leigh B. Stoller's avatar
      Another CMU hack; Add a "remote" mode (-m) that tells capture to connect to · 54a55729
      Leigh B. Stoller authored
      a socket on a remote node instead of a local tty device. When using
      remotemode, the last argument to capture should be host:port instead
      of tty.
    • Leigh B. Stoller's avatar
      Another change for CMU; they have laptops that do not power up after a · ec9ed5f7
      Leigh B. Stoller authored
      power cycle. For those we need to send wakeonlan packets to kick them.
      This is complicated by the fact that the laptops take an incredibly
      long time to go from poweron to a point where it will listen to a
      wakeonlan packet. No idea why. I use StateWait() to find out when the
      node hits bootinfo (which sends a PXEBOOTING event). Keep looping
      sending more wakeonlan packets until we hear from the node. Time out
      the operation after two minutes (probably need something to control
      To find the outgoing interface required by tools/whol program, I look
      in the interfces table for boss' control network.
      The mac address of the target node comes from the interfaces table.
      The flag that controls whether a node needs wakeonlan is in the
      node_attributes table:
      	mysql> insert into node_attributes set
      		  attrvalue=1, node_id='pcxxx';
  9. 25 Feb, 2008 3 commits
  10. 22 Feb, 2008 3 commits
  11. 21 Feb, 2008 4 commits
  12. 20 Feb, 2008 3 commits
  13. 19 Feb, 2008 1 commit
  14. 18 Feb, 2008 2 commits
    • Leigh B. Stoller's avatar
      Fix cvs merge error. · 2c7555bc
      Leigh B. Stoller authored
    • Leigh B. Stoller's avatar
      Add support for hashified/cloned filesystems. This is really only · 4bb9e149
      Leigh B. Stoller authored
      useful on widearea nodes where we can amortize the time it takes to
      copy/hash the necessary filesystems, which takes about 15 minutes.
      We would do this when a node first boots (or its image is updated I
      Note that the clone operation (when a vserver is actually created) is very fast
      though, taking about 30 seconds.
      Currently controlled by local variable $USEHASHIFIED for debugging.
  15. 16 Feb, 2008 1 commit
  16. 15 Feb, 2008 1 commit
  17. 14 Feb, 2008 3 commits
  18. 13 Feb, 2008 3 commits