...and also describe the general approach of auto-balancing by the maintenance daemon.
Signed-off-by: Klaus Aehlig <[email protected]> --- man/gnt-cluster.rst | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/man/gnt-cluster.rst b/man/gnt-cluster.rst index 6ce43b1..f95328f 100644 --- a/man/gnt-cluster.rst +++ b/man/gnt-cluster.rst @@ -713,6 +713,8 @@ MODIFY | [\--enabled-data-collectors *collectors*] | [\--data-collector-interval *intervals*] | [\--maintenance-interval *seconds*] +| [\--auto-balance-cluster {yes \| no }] +| [\--auto-balance-threshold *score* ] Modify the options for the cluster. @@ -786,6 +788,13 @@ shall be collected. The ``--maintenance-interval`` option specified the minimal waiting time by the maintenance daemon between maintenance rounds. +The ``--auto-balance-cluster`` option tell the maintenance daemon +whether to also keep the cluster in a balanced fashion. If so, it +will carry out moves, provided the gain in the cluster score for +that move is at least the value specified by ``--auto-balance-threshold`` +in absolute terms, unless the cluster score it at least 10 times that +value, in which case all beneficial steps will be done if auto-balancing +is enabled. See **gnt-cluster init** for a description of ``--install-image`` and ``--zeroing-image``. -- 2.4.3.573.g4eafbef
