ndb_index_stat provides per-fragment statistical information about indexes on NDB tables. This includes cache version and age, number of index entries per partition, and memory consumption by indexes.

Usage

To obtain basic index statistics about a given NDB table, invoke ndb_index_stat as shown here, with the name of the table as the first argument and the name of the database containing this table specified immediately following it, using the --database (-d) option:

ndb_index_stat table -d database

In this example, we use ndb_index_stat to obtain such information about an NDB table named mytable in the test database:

shell> ndb_index_stat -d test mytable
table:City index:PRIMARY fragCount:2
sampleVersion:3 loadTime:1399585986 sampleCount:1994 keyBytes:7976
query cache: valid:1 sampleCount:1994 totalBytes:27916
times in ms: save: 7.133 sort: 1.974 sort per sample: 0.000

NDBT_ProgramExit: 0 - OK

sampleVersion is the version number of the cache from which the statistics data is taken. Running ndb_index_stat with the --update option causes sampleVersion to be incremented.

loadTime shows when the cache was last updated. This is expressed as seconds since the Unix Epoch.

sampleCount is the number of index entries found per partition. You can estimate the total number of entries by multiplying this by the number of fragments (shown as fragCount).

sampleCount can be compared with the cardinality of SHOW INDEX or INFORMATION_SCHEMA.STATISTICS, although the latter two provide a view of the table as a whole, while ndb_index_stat provides a per-fragment average.

keyBytes is the number of bytes used by the index. In this example, the primary key is an integer, which requires four bytes for each index, so keyBytes can be calculated in this case as shown here:

    keyBytes = sampleCount * (4 bytes per index) = 1994 * 4 = 7976

This information can also be obtained using the corresponding column definitions from INFORMATION_SCHEMA.COLUMNS (this requires a MySQL Server and a MySQL client application).

totalBytes is the total memory consumed by all indexes on the table, in bytes.

Timings shown in the preceding examples are specific to each invocation of ndb_index_stat.

The --verbose option provides some additional output, as shown here:

shell> ndb_index_stat -d test mytable --verbose
random seed 1337010518
connected
loop 1 of 1
table:mytable index:PRIMARY fragCount:4
sampleVersion:2 loadTime:1336751773 sampleCount:0 keyBytes:0
read stats
query cache created
query cache: valid:1 sampleCount:0 totalBytes:0
times in ms: save: 20.766 sort: 0.001
disconnected

NDBT_ProgramExit: 0 - OK

shell>

Options

The following table includes options that are specific to the NDB Cluster ndb_index_stat utility. Additional descriptions are listed following the table. For options common to most NDB Cluster programs (including ndb_index_stat), see Section 18.4.26, “Options Common to NDB Cluster Programs — Options Common to NDB Cluster Programs”.

ndb_index_stat statistics options.  The following options are used to generate index statistics. They work with a given table and database. They cannot be mixed with system options (see ndb_index_stat system options).

ndb_index_stat system options.  The following options are used to generate and update the statistics tables in the NDB kernel. None of these options can be mixed with statistics options (see ndb_index_stat statistics options).