This topic contains an example of Eneperf output and describes it briefly.
Running iteration 1... Done: 58881 sent, 58881 received, 0 errors. 22 minutes, 42.63 seconds for iteration 1, 43.2112 req/sec. 22 minutes, 42.63 seconds elapsed (user: 6.20 seconds, system: 15.24 seconds). Net: 1.18389e+06 KB (868.829 KB/sec). Page Size: avg=91.34 KB, std dev=142.81 KB, max=1238.37 KB, min=0.16 KB. Latency: avg=92.36 ms, std dev=238.27 ms, max=13441.11 ms, min=0.18 ms. 250 queries longer than 1s. Eneperf completed: 58881 sent, 58881 received, 0 errors. 22 minutes, 42.63 seconds elapsed (user: 6.20 seconds, system: 15.24 seconds). Net: 1.18389e+06 KB (868.829 KB/sec). Page Size: avg=91.34 KB, std dev=142.81 KB, max=1238.37 KB, min=0.16 KB. Latency: avg=92.36 ms, std dev=238.27 ms, max=13441.11 ms, min=0.18 ms. 250 queries longer than 1s. Best iteration time: 22 minutes, 42.63 seconds. Peak rate: 43.2112 req/sec. Avg iteration time: 22 minutes, 42.63 seconds. Avg rate: 43.2112 req/sec. Total rate: 43.2112 req/sec.
The entries from Eneperf output are described in the following table:
Sample Eneperf output entry |
Description |
---|---|
|
Is printed as each iteration begins. The numbers following this line, until "Eneperf completed:" occur for each iteration requested. The number of iterations requested is the last Eneperf parameter. |
|
Is printed once the iteration finishes. |
|
“Sent” is the number of queries sent. It is the sum of “Received” and “Errors” and the number of errors, where errors is the number of 404 or 400 HTTP codes that the Dgraph returns, (rather than errors in the Dgraph log). “Received” is the number of queries with a 200 HTTP status code that the Dgraph returns. “Errors” is the number of queries with 404 or 400 HTTP status code that the Dgraph returns, rather than errors in the Dgraph log. |
|
The time for the specific iteration, and the throughput for this iteration. |
|
The total runtime up until this point. System time is the time spent in the operating system on behalf of the Dgraph. User time is the time spent in the Dgraph itself. |
|
The total amount of data returned for the entire test (not just for one iteration). |
|
Cumulative statistics on the amount of data returned for each query. |
|
Cumulative statistics on the latencies. The statistics include previous iterations. Latency information may be inaccurate when multiple connections are in use, particularly if the network is slow. If accuracy is critical, consider obtaining latency information from the Dgraph request log. |
|
The processing rate of the iteration in the test with the best performance, but should not be confused with "peak" performance in the sense of a single second that showed the highest throughput. It is the total number of requests processed in that iteration divided by the time of the iteration in seconds. If the test includes only one iteration, peak rate is the processing rate for that iteration. |
|
The average time of the iterations in the test. |
|
The average rate of the iterations in the test, in requests processed per second. |
|
The total of requests processed for all of the iterations in the test, divided by the total time of all of the iterations in the test. |
|
All information after this statement is cumulative over the entire run. This line is printed once all iterations have completed. |