Write the time taken by each transaction to a log file.

Length of aggregation interval (in seconds). May be used only together

with <application>-l</application> - with this option, the log contains

per-interval summary (number of transactions, min/max latency and two

additional fields useful for variance estimation).

</para>

<para>

This option is not currently supported on Windows.

Set the filename prefix for the transaction log file created by

<option>--log</>. The default is <replaceable>pgbench_log</>.

With the <option>-l</> option but without the <option>--aggregate-interval</option>,

<application>pgbench</> writes the time taken by each transaction

<application>pgbench</application> process. If the <option>-j</> option is 2 or higher,

creating multiple worker threads, each will have its own log file. The first worker will

<filename><replaceable>pgbench_log</>.<replaceable>nnn</>.<replaceable>mmm</></filename>,

with 1. The prefix can be changed by using the <option>--log-prefix</>

option.

<replaceable>client_id</> <replaceable>transaction_no</> <replaceable>time</> <replaceable>script_no</> <replaceable>time_epoch</> <replaceable>time_us</> <optional><replaceable>schedule_lag</replaceable></optional>

where <replaceable>time</> is the total elapsed transaction time in microseconds,

Unix epoch format time stamp and an offset

Field <replaceable>schedule_lag</> is the difference between the

Here is a snippet of the log file generated:

Another example with --rate=100 and --latency-limit=5 (note the additional

With the <option>--aggregate-interval</option> option, the logs use a bit different format:

<replaceable>interval_start</> <replaceable>num_of_transactions</> <replaceable>latency_sum</> <replaceable>latency_2_sum</> <replaceable>min_latency</> <replaceable>max_latency</> <optional><replaceable>lag_sum</> <replaceable>lag_2_sum</> <replaceable>min_lag</> <replaceable>max_lag</> <optional><replaceable>skipped_transactions</></optional></optional>

where <replaceable>interval_start</> is the start of the interval (Unix epoch

format time stamp), <replaceable>num_of_transactions</> is the number of transactions

within the interval, <replaceable>latency_sum</replaceable> is a sum of latencies

(so you can compute average latency easily). The following two fields are useful

for variance estimation - <replaceable>latency_sum</> is a sum of latencies and

<replaceable>latency_2_sum</> is a sum of 2nd powers of latencies. The next two

fields are <replaceable>min_latency</> - a minimum latency within the interval, and

<replaceable>max_latency</> - maximum latency within the interval. A transaction is

counted into the interval when it was committed. The fields in the end,

<replaceable>lag_sum</>, <replaceable>lag_2_sum</>, <replaceable>min_lag</>,

option is used. The very last one, <replaceable>skipped_transactions</>,

is only present if the option <option>--latency-limit</> is present, too.

They are calculated from the time each transaction had to wait for the

Here is example output:

Notice that while the plain (unaggregated) log file contains a reference

to the custom script files, the aggregated log does not. Therefore if

you need per script data, you need to aggregate the data on your own.

long start_time; /* interval start time, for aggregates */

static void doLog(TState *thread, CState *st, instr_time *now,

initStats(StatsData *sd, double start_time)

doLog(TState *thread, CState *st, instr_time *now,

while (agg->start_time + agg_interval < INSTR_TIME_GET_DOUBLE(*now))

agg->start_time,

/*

if ((!skipped || agg_interval) && INSTR_TIME_IS_ZERO(*now))

doLog(thread, st, now, agg, skipped, latency, lag);

initStats(&ps.stats, 0.0);

#ifdef WIN32

fprintf(stderr, "--aggregate-interval is not currently supported on Windows\n");

exit(1);

initStats(&thread->stats, 0.0);

initStats(&stats, 0.0);

initStats(&aggs, INSTR_TIME_GET_DOUBLE(thread->start_time));

last = aggs;

initStats(&cur, 0.0);

if (agg_interval)

doLog(thread, state, &end, &aggs, false, 0, 0);