All compaction related settings in the yaml were untouched. The fingerprint column family has been populated three days ago and the cpu/disk usage were pretty low. I'd think Cassandra will silently start the compaction thread on my behalf and try to preserve the min/max thresholds, rather than waiting for a major compaction order from nodetool explicitly.
Anyway, I did a major compaction on all 5 nodes almost at the same time. 4 of them came back after a few hours, but one of the 5 nodes still has a lot of pending ones: cassandra/bin/nodetool -h localhost compactionstats pending tasks: 76 And the uptime is very light. 14:31:44 up 30 days, 22:11, 4 users, load average: 0.29, 0.58, 0.58 Some reply suggested this is a broken compaction. I will wait for a few hours and restart that node if nothing changes. -- View this message in context: http://cassandra-user-incubator-apache-org.3065146.n2.nabble.com/Read-latency-is-over-1-minute-on-a-column-family-with-400-000-rows-tp6639649p6642279.html Sent from the cassandra-u...@incubator.apache.org mailing list archive at Nabble.com.
