Skip to content
Snippets Groups Projects
  1. Jun 28, 2011
  2. Jun 24, 2011
  3. Jun 23, 2011
  4. Jun 22, 2011
  5. Jun 17, 2011
  6. Jun 15, 2011
  7. Jun 14, 2011
  8. Jun 10, 2011
  9. Jun 09, 2011
  10. Jun 08, 2011
    • Michael Hanselmann's avatar
      Fix locking issues in LUClusterVerifyGroup · fceb01fe
      Michael Hanselmann authored
      
      - Use functions in ConfigWriter instead of custom loops
      - Calculate nodes only once instances locks are acquired, removes one
        potential race condition
      - Don't retrieve lists of all node/instance information without locks
      - Additionally move the end of the node time check window after the
        first RPC call--the second call isn't involved in checking the
        node time at all
      
      Signed-off-by: default avatarMichael Hanselmann <hansmi@google.com>
      Reviewed-by: default avatarRené Nussbaumer <rn@google.com>
      fceb01fe
    • Michael Hanselmann's avatar
      cmdlib: Acquire BGL for LUClusterVerifyConfig · c5312a10
      Michael Hanselmann authored
      
      LUClusterVerifyConfig verifies a number of configuration settings. For
      doing so, it needs a consistent list of nodes, groups and instances. So
      far no locks were acquired at all (except for the BGL in shared mode).
      This is a race condition (e.g. if a node group is added in parallel) and
      can be fixed by acquiring the BGL in exclusive mode. Since this LU
      verifies the cluster-wide configuration, doing so instead of acquiring
      individual locks is just.
      
      Includes one typo fix and one docstring update.
      
      Signed-off-by: default avatarMichael Hanselmann <hansmi@google.com>
      Reviewed-by: default avatarRené Nussbaumer <rn@google.com>
      c5312a10
  11. Jun 07, 2011
  12. Jun 03, 2011
  13. Jun 01, 2011
Loading