Revision 0102e732

b/doc/design-hroller.rst
123 123
managing better RBD pools, if those are implemented on top of nodes
124 124
storage, rather than on dedicated storage machines.
125 125

  
126
Full-Evacuation
127
+++++++++++++++
128

  
129
If full evacuation of the nodes to be rebooted is desired, a simple
130
migration is not enough for the DRBD instances. To keep the number of
131
disk operations small, we restrict moves to ``migrate, replace-secondary``.
132
That is, after migrating instances out of the nodes to be rebooted,
133
replacement secondaries are searched for, for all instances that have
134
their then secondary on one of the rebooted nodes. This is done by a
135
greedy algorithm, refining the initial reboot partition, if necessary.
136

  
126 137
Future work
127 138
===========
128 139

  

Also available in: Unified diff