One more note, When we did this conversion, we were on Cassandra 1.1.X. You didn't mention what version of Cassandra you were running,
Thanks, -Mike On Oct 23, 2013, at 10:05 AM, Michael Theroux wrote: > When we made a similar move, for an unknown reason (I didn't hear any > feedback from the list when I asked why this might be), compaction didn't > start after we moved from SizedTiered to leveled compaction until I ran > "nodetool compact <keyspace> <column-family-converted-to-lcs>". > > The thread is here: > > http://www.mail-archive.com/user@cassandra.apache.org/msg27726.html > > I've also seen other individuals on this list state that those pending > compaction stats didn't move unless the node was restarted. Compaction > started to run several minutes after restart. > > Thanks, > -Mike > > On Oct 23, 2013, at 9:14 AM, Russ Garrett wrote: > >> Hi, >> >> We have a cluster which we've recently moved to use >> LeveledCompactionStrategy. We were experiencing some disk space >> issues, so we added two additional nodes temporarily to aid >> compaction. Once the compaction had completed on all nodes, we >> decommissioned the two temporary nodes. >> >> All nodes now have a high number of pending tasks which isn't dropping >> - they're remaining approximately static. There are constantly >> compaction tasks running, but when they complete, the pending tasks >> number doesn't drop. We've set the compaction rate limit to 0, and >> increased the number of compactor threads until the I/O utilisation is >> at maximum, but neither of these have helped. >> >> Any suggestions? >> >> Cheers, >> >> -- >> Russ Garrett >> r...@garrett.co.uk >
