Database Reference
In-Depth Information
The columns shown are address, status, state, load, ownership percentage, and
tokens. These are all self-explanatory. What you are looking at is the current state
of the ring as seen by the querying node.
A few things are worth noting as you look at the output. Notice that the load
on each machine is roughly the same, as each owns 33.33% of the data in the ring.
The size on disk is different because different data is stored on each node. Addi-
tionally, every row must fully fit on one node. If you have wide rows, the more
data you get, the more you will see the size in the load section vary from node to
node. Last, notice that the last token on the bottom is also the same token listed
above the first node. This is the way the wrapping range is displayed.
Since the status on all three nodes is up and the state on all three nodes is nor-
mal, we can consider this a healthy ring with some healthy nodes and move on.
ColumnFamily Statistics
In addition to monitoring the health of a node or your cluster, you can also gather
some pretty granular statistics using
nodetool
. On the ColumnFamily level, you
can run
nodetool cfstats
and see statistics for your cluster at the keyspace
and ColumnFamily levels.
Note that you may want to pipe the output through
less
or send the output to a
file. Only the keyspace statistics and ColumnFamily statistics are included in
List-
ing 7.5
for a single ColumnFamily in a single keyspace. Each keyspace (including
the System Keyspace) and every ColumnFamily within each keyspace will nor-
mally be output when this command is run.
Listing 7.5
Example ColumnFamily Stats Output
$ nodetool cfstats
Keyspace: MainKeyspace
Read Count: 2032889905
Read Latency: 2.2192086369876485 ms.
Write Count: 6288511536
Write Latency: 0.03336843706761708 ms.
Pending Tasks: 0
Column Family: events_2013_01
SSTable count: 12
