I should note that these OSDs also drop out of the pool as part of their symptoms - it's not clear to me at the moment if they drop out *because* of the memory, or if the buffer_pool is growing large because it's buffering communications that aren't getting to the cluster [and hence they drop out]. _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |