So basically this means that {tables: rows}, for example, {1: 3, 3: 1} means that three rows of 3 (3: 1) are taken from sstable (1: 3) and 1 row sstables, all to make one failure in the compaction operation.
I tried this myself, here is an example, hope this helps:
create a keyspace and table:
cqlsh> create keyspace space1 WITH replication = {'class': 'SimpleStrategy', 'replication_factor': 1}; cqlsh> create TABLE space1.tb1 ( key text, val1 text, primary KEY (key)); cqlsh> INSERT INTO space1.tb1 (key, val1 ) VALUES ( 'key1','111'); cqlsh> INSERT INTO space1.tb1 (key, val1 ) VALUES ( 'key2','222'); cqlsh> INSERT INTO space1.tb1 (key, val1 ) VALUES ( 'key3','333'); cqlsh> INSERT INTO space1.tb1 (key, val1 ) VALUES ( 'key4','444'); cqlsh> INSERT INTO space1.tb1 (key, val1 ) VALUES ( 'key5','555'); cqlsh> exit
Now we hide to create sstable
$ nodetool flush space1
We see that only one version of the table is created
$ sudo ls -lR /var/lib/cassandra/data/space1 /var/lib/cassandra/data/space1: total 4 drwxr-xr-x. 2 cassandra cassandra 4096 Feb 3 12:51 tb1 /var/lib/cassandra/data/space1/tb1: total 32 -rw-r--r--. 1 cassandra cassandra 43 Feb 3 12:51 space1-tb1-jb-1-CompressionInfo.db -rw-r--r--. 1 cassandra cassandra 146 Feb 3 12:51 space1-tb1-jb-1-Data.db -rw-r--r--. 1 cassandra cassandra 24 Feb 3 12:51 space1-tb1-jb-1-Filter.db -rw-r--r--. 1 cassandra cassandra 90 Feb 3 12:51 space1-tb1-jb-1-Index.db -rw-r--r--. 1 cassandra cassandra 4389 Feb 3 12:51 space1-tb1-jb-1-Statistics.db -rw-r--r--. 1 cassandra cassandra 80 Feb 3 12:51 space1-tb1-jb-1-Summary.db -rw-r--r--. 1 cassandra cassandra 79 Feb 3 12:51 space1-tb1-jb-1-TOC.txt
check sstable2json, we see our data
$ sudo -u cassandra /usr/bin/sstable2json /var/lib/cassandra/data/space1/tb1/space1-tb1-jb-1-Data.db [ {"key": "6b657935","columns": [["","",1422967847005000], ["val1","555",1422967847005000]]}, {"key": "6b657931","columns": [["","",1422967817740000], ["val1","111",1422967817740000]]}, {"key": "6b657934","columns": [["","",1422967840622000], ["val1","444",1422967840622000]]}, {"key": "6b657933","columns": [["","",1422967832341000], ["val1","333",1422967832341000]]}, {"key": "6b657932","columns": [["","",1422967825116000], ["val1","222",1422967825116000]]} ]
At this point, 'notetool compactionhistory shows nothing for this table, but allows you to run a compact file in any case, to see what we get (scroll to the right)
$ nodetool compactionhistory | awk 'NR == 2 || /space1/' id keyspace_name columnfamily_name compacted_at bytes_in bytes_out rows_merged 5725f890-aba4-11e4-9f73-351725b0ac5b space1 tb1 1422968305305 146 146 {1:5}
Now let's delete two lines and discard
cqlsh> delete from space1.tb1 where key='key1'; cqlsh> delete from space1.tb1 where key='key2'; cqlsh> exit $ nodetool flush space1 $ sudo ls -l /var/lib/cassandra/data/space1/tb1/ [sudo] password for datastax: total 64 -rw-r--r--. 1 cassandra cassandra 43 Feb 3 12:58 space1-tb1-jb-2-CompressionInfo.db -rw-r--r--. 1 cassandra cassandra 146 Feb 3 12:58 space1-tb1-jb-2-Data.db -rw-r--r--. 1 cassandra cassandra 336 Feb 3 12:58 space1-tb1-jb-2-Filter.db -rw-r--r--. 1 cassandra cassandra 90 Feb 3 12:58 space1-tb1-jb-2-Index.db -rw-r--r--. 1 cassandra cassandra 4393 Feb 3 12:58 space1-tb1-jb-2-Statistics.db -rw-r--r--. 1 cassandra cassandra 80 Feb 3 12:58 space1-tb1-jb-2-Summary.db -rw-r--r--. 1 cassandra cassandra 79 Feb 3 12:58 space1-tb1-jb-2-TOC.txt -rw-r--r--. 1 cassandra cassandra 43 Feb 3 13:02 space1-tb1-jb-3-CompressionInfo.db -rw-r--r--. 1 cassandra cassandra 49 Feb 3 13:02 space1-tb1-jb-3-Data.db -rw-r--r--. 1 cassandra cassandra 16 Feb 3 13:02 space1-tb1-jb-3-Filter.db -rw-r--r--. 1 cassandra cassandra 36 Feb 3 13:02 space1-tb1-jb-3-Index.db -rw-r--r--. 1 cassandra cassandra 4413 Feb 3 13:02 space1-tb1-jb-3-Statistics.db -rw-r--r--. 1 cassandra cassandra 80 Feb 3 13:02 space1-tb1-jb-3-Summary.db -rw-r--r--. 1 cassandra cassandra 79 Feb 3 13:02 space1-tb1-jb-3-TOC.txt
Allows you to check the contents of tables
$ sudo -u cassandra /usr/bin/sstable2json /var/lib/cassandra/data/space1/tb1/space1-tb1-jb-2-Data.db [ {"key": "6b657935","columns": [["","",1422967847005000], ["val1","555",1422967847005000]]}, {"key": "6b657931","columns": [["","",1422967817740000], ["val1","111",1422967817740000]]}, {"key": "6b657934","columns": [["","",1422967840622000], ["val1","444",1422967840622000]]}, {"key": "6b657933","columns": [["","",1422967832341000], ["val1","333",1422967832341000]]}, {"key": "6b657932","columns": [["","",1422967825116000], ["val1","222",1422967825116000]]} ] $ sudo -u cassandra /usr/bin/sstable2json /var/lib/cassandra/data/space1/tb1/space1-tb1-jb-3-Data.db [ {"key": "6b657931","metadata": {"deletionInfo": {"markedForDeleteAt":1422968551313000,"localDeletionTime":1422968551}},"columns": []}, {"key": "6b657932","metadata": {"deletionInfo": {"markedForDeleteAt":1422968553322000,"localDeletionTime":1422968553}},"columns": []} ]
Now let's compact
$ nodetool compact space1
Only one stable as expected
$ sudo ls -l /var/lib/cassandra/data/space1/tb1/ total 32 -rw-r--r--. 1 cassandra cassandra 43 Feb 3 13:05 space1-tb1-jb-4-CompressionInfo.db -rw-r--r--. 1 cassandra cassandra 133 Feb 3 13:05 space1-tb1-jb-4-Data.db -rw-r--r--. 1 cassandra cassandra 656 Feb 3 13:05 space1-tb1-jb-4-Filter.db -rw-r--r--. 1 cassandra cassandra 90 Feb 3 13:05 space1-tb1-jb-4-Index.db -rw-r--r--. 1 cassandra cassandra 4429 Feb 3 13:05 space1-tb1-jb-4-Statistics.db -rw-r--r--. 1 cassandra cassandra 80 Feb 3 13:05 space1-tb1-jb-4-Summary.db -rw-r--r--. 1 cassandra cassandra 79 Feb 3 13:05 space1-tb1-jb-4-TOC.txt
Now let's check the contents of the new stable, we will see tombstones
$ sudo -u cassandra /usr/bin/sstable2json /var/lib/cassandra/data/space1/tb1/space1-tb1-jb-4-Data.db [ {"key": "6b657935","columns": [["","",1422967847005000], ["val1","555",1422967847005000]]}, {"key": "6b657931","metadata": {"deletionInfo": {"markedForDeleteAt":1422968551313000,"localDeletionTime":1422968551}},"columns": []}, {"key": "6b657934","columns": [["","",1422967840622000], ["val1","444",1422967840622000]]}, {"key": "6b657933","columns": [["","",1422967832341000], ["val1","333",1422967832341000]]}, {"key": "6b657932","metadata": {"deletionInfo": {"markedForDeleteAt":1422968553322000,"localDeletionTime":1422968553}},"columns": []} ]
Finally, let's check the compaction history (scroll right)
$ nodetool compactionhistory | awk 'NR == 2 || /space1/' id keyspace_name columnfamily_name compacted_at bytes_in bytes_out rows_merged 5725f890-aba4-11e4-9f73-351725b0ac5b space1 tb1 1422968305305 146 146 {1:5} 46112600-aba5-11e4-9f73-351725b0ac5b space1 tb1 1422968706144 195 133 {1:3, 2:2}