cluster with 4 nodes --copies 3. Kill node 4 and wait the recovery to end. Kill node 3, Cluster status: IO has halted as there are too few living nodes. In my head :-) the cluster is "frozen". Also 'node recovery' gives empty output. Why is node 3 going the be recovered once back? Nodes In Recovery: Id Host:Port V-Nodes Zone 2 192.168.2.43:7000 92 721594560 sheep.log ... Feb 13 11:46:39 [rw 443] recover_object_work(201) done:439 count:444, oid:a34c6700000634 Feb 13 11:46:39 [rw 444] recover_object_work(201) done:440 count:444, oid:a34c67000004d6 Feb 13 11:46:39 [rw 445] recover_object_work(201) done:441 count:444, oid:a34c67000005a0 Feb 13 11:46:39 [rw 446] recover_object_work(201) done:442 count:444, oid:a34c670000048d Feb 13 11:46:39 [rw 447] recover_object_work(201) done:443 count:444, oid:a34c67000004ec Feb 13 11:46:39 [main] queue_cluster_request(307) COMPLETE_RECOVERY (0xe3bc40) It seems like it's overwriting the chunks. Is it possible to understand from sheep.log if chunks are overwritten/added or "marked as good"? |