Allow unconditional failovers off offline nodes
authorKlaus Aehlig <aehlig@google.com>
Fri, 29 May 2015 15:50:43 +0000 (17:50 +0200)
committerKlaus Aehlig <aehlig@google.com>
Fri, 29 May 2015 16:18:39 +0000 (18:18 +0200)
commit4ca31587850b8d60b12f3be379cc0186387f2895
treec5ff0ce9e124a64986c11855f867fcd7da53071e
parent47848c229ef48884fbb62e581c024bbf0f5110aa
Allow unconditional failovers off offline nodes

Normally, we should not place instances on nodes that do
not have enough disks. However, there is one exception: if
we failover an instance from an offline node, that node can
well be secondary of that instance---the fact that it is currently
primary proves it has enough disks space. The reason why we have
to handle that case special is that if a node is offline, we sometimes
cannot determine the amount of disk available, hence the conservative
estimation is 0.

Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Hrvoje Ribicic <riba@google.com>
src/Ganeti/HTools/Cluster.hs
src/Ganeti/HTools/Node.hs