From: Izhar <[email protected]>

---
 doc/design-hugepages-ganeti-support.rst | 26 +++++++++++++++++---------
 1 file changed, 17 insertions(+), 9 deletions(-)

diff --git a/doc/design-hugepages-ganeti-support.rst 
b/doc/design-hugepages-ganeti-support.rst
index a4ba827..ef5ae6e 100644
--- a/doc/design-hugepages-ganeti-support.rst
+++ b/doc/design-hugepages-ganeti-support.rst
@@ -34,21 +34,26 @@ cluster level via the hypervisor parameter ``mem_path`` as::
        >--enabled-hypervisors=kvm -nic-parameters link=br100 \
        > -H kvm:mem_path=/mount/point/for/hugepages
 
-This hypervisor parameter is  inherited by all the instances as
+This hypervisor parameter is inherited by all the instances as
 default although it can be overriden at the instance level. 
 
 The following changes will be made to the inheritence behaviour.
 
--  The hypervisor parameter ``mem_path`` must be made available at the
-   node group level (in addition to the cluster level),so that users can
-   set it as default for the node group::
+-  The hypervisor parameter   ``mem_path`` and all other hypervisor
+   parameters will be made available at the node group level (in
+   addition to the cluster level), so that users can set defaults for
+   the node group::
 
        $ gnt-group add/modify\
-       > -H kvm:mem_path=/mount/point/for/hugepages
+       > -H hv:parameter=value
+
+   This changes the hypervisor inheritence level as::
+       
+       cluster -> group -> OS -> instance
 
 -  Furthermore, the hypervisor parameter ``mem_path`` will be changeable
    only at the cluster or node group level and users must not be able to
-   override this when creating new instances. The following command must
+   override this at OS or instance level. The following command must
    produce an error message that ``mem_path`` may only be set at either
    the cluster or the node group level::
 
@@ -66,18 +71,21 @@ the memory reserved for huge pages as:
 - mpgsize/hp (Size of a huge page)
 
 mfree and mtotal will be changed to mean "the total and free memory for
-the default method in this cluster/nodegroup"
+the default method in this cluster/nodegroup". Note that the default
+method depends both on the default hypervisor and its parameters.
 
 iAllocator Changes:
 -------------------
 If huge pages are set as default for a cluster of node group, then
 iAllocator must consider the huge pages memory on the nodes, as a
 parameter when trying to find the best node for the VM.
+Note that the iallocator will also be changed to use the correct
+parameter depending on the cluster/group.
 
 hbal Changes:
 -------------
-The cluster balancer (hbal) will also be changed to consider memory
-pools and recognize memory reserved for huge pages when trying to
+The cluster balancer (hbal) will be changed to use the default  memory
+pool and  recognize memory reserved for huge pages when trying to
 rebalance the cluster.
 
 .. vim: set textwidth=72 :
-- 
1.8.2.1

Reply via email to