Went through all the patches again - LGTM for all!
On Wed, Jun 25, 2014 at 2:33 PM, 'Klaus Aehlig' via ganeti-devel < [email protected]> wrote: > > Generally, it is a feature to be conservative and estimate the > capacity of a cluster to 0 if there is at least one node not N+1 > happy. However, there are situation where this estimate is too > conserevative, especially in multi--node-group clusters. Add two > options that support more accurate estimations: one will allow > to consider the different node groups independently, i.e., allocate > on all N+1 happy node groups, and the second allows to consider > allocation on all N+1 happy nodes. > > Fixes issue 861. > > Klaus Aehlig (10): > hspace: optionally print allocatable nodes > Add a modifier for a group's allocation policy > Add an option --independent-groups > Make hspace support --independent-groups > Verify independence of allocation > Document --independent-groups > Add an option to accept existing N+1 violations > hspace: support --accept-existing-errors > Test hspace --accept-existing > Document --accept-existing-errors > > Makefile.am | 2 ++ > man/hspace.rst | 14 ++++++++++++++ > src/Ganeti/HTools/CLI.hs | 20 ++++++++++++++++++++ > src/Ganeti/HTools/Group.hs | 5 +++++ > src/Ganeti/HTools/Program/Hspace.hs | 23 ++++++++++++++++++++--- > test/data/htools/hspace-groups-one.data | 14 ++++++++++++++ > test/data/htools/hspace-groups-two.data | 16 ++++++++++++++++ > test/hs/shelltests/htools-hspace.test | 26 ++++++++++++++++++++++++++ > 8 files changed, 117 insertions(+), 3 deletions(-) > create mode 100644 test/data/htools/hspace-groups-one.data > create mode 100644 test/data/htools/hspace-groups-two.data > > -- > 2.0.0.526.g5318336 > >
