The tc_time_track_stats table provides time-tracking information obtained from the DBTC block (TC) instances in the data nodes, through API nodes access NDB. Each TC instance tracks latencies for a set of activities it undertakes on behalf of API nodes or other data nodes; these activities include transactions, transaction errors, key reads, key writes, unique index operations, failed key operations of any type, scans, failed scans, fragment scans, and failed fragment scans.

A set of counters is maintained for each activity, each counter covering a range of latencies less than or equal to an upper bound. At the conclusion of each activity, its latency is determined and the appropriate counter incremented. tc_time_track_stats presents this information as rows, with a row for each instance of the following:

Each row contains a value for each activity type. This is the number of times that this activity occurred with a latency within the range specified by the row (that is, where the latency does not exceed the upper bound).

The following table provides information about the columns in tc_time_track_stats. For each column, the table shows the name, data type, and a brief description. Additional information can be found in the notes following the table.

The block_instance column provides the DBTC kernel block instance number. You can use this together with the block name to obtain information about specific threads from the threadblocks table.