Ceph

In reference to #272 i create this issue.

During this replication change i had some crashes at the OSD's. On almost all the machines i got messages that the "cosd" process was "hung" and "blocking", but some also started to segfault.

I gathered some core dumps and uploaded them at: http://zooi.widodh.nl/ceph/core_dumps_osd_crash_replication_change.tar

The problem is, there are no logs, don't know what happened, but it seems the logs have been truncated in some way. There was enough diskspace for the logs.

I'm going to reproduce the whole situation, in order to get good logfiles this time.

But maybe there is some valuable information in these coredumps, i don't know.